Interview Questions on digital signal processing
1. If x[n] is a N point sequence, what is the difference between N point DFT and N+1 point DFT.
2. How to find the convolution of two sequences using DFT operation
3. What is the difference between DFT and DTFT. What is the relation between them.
4. What is the need for DFT when DTFT is already present
5. How to find the dft of two N point real sequences using a N point DFT
6. What is the reduction in computational complexity by implementing DFT as FFT
7. If you have an N point dft calculator and two N/2 sequences. How can you find the dft of the sequences
8. How to find the dft of a 2N point real sequence with a N point DFT hardware
9. What are overlap save and overlap add methods and what are they used for
10. If we limit a signal to L samples, what is the maximum difference between frequencies that will be seen as same frequency.
11. Describe the process of analog to digital conversion and the effect of quantization noise on resulting SNR (where noise is quantization noise)
12. What is the application of Cross correlation and Auto Correlation
13. If you have a signal x[n] and it is upsampled by N and then downsampled by M. what is the condition for not losing any information in x[n]
14. How do u reduce spectral leakage
15. What is the speciality about minimum phase filter
16. why IIR filters doesn't have Linear phase
17. Can IIR filters be Linear phase? how to make it linear Phase?
18. Describe the process of decimation. what is the cutoff frequency of the filter to be used and why is it required
19. Describe the process of Interpolation. what is the cutoff frequency of the filter to be used and why is it required
20. What is the simplest high pass filter. Write the equation
21. What is the difference between equiripple filter and FIR filter
22. What’s differences b/w butter worth, chebyshev, elliptical filter and advantages/disadvantages of each
23. why we use DCT extensively in compression
24. Why after DCT we use a zig zag manner for run length coding
25. What is the basic difference between FIR and IIR filters
26. What are limit cycle oscillations
27. Is the system y = ax + n linear
28. If you have 16 bit multiplier hardware, how do you find 32 bit multiplication
29. If there is a discontinuity in the signal amplitude in time, why will there be more spectral components
30. If we have a 10 bit input sample and an FIR filter with 15 taps each of 6 bits, what is the minimum number of bits required to store the output
31. What is the importance of linear phase of a filter and it’s relation to group delay
32. How does the location of poles and their order affect the stability of a system
33. What is a continuous time signal, discrete signal and digital signal
34. Describe discrete time sinusoid signal. What is the condition for the signal to be periodic
35. How are signals classified into energy and power signals
36. What are causal systems
37. When is a system said to be stable
38. What is the condition for stability of linear time invariant systems
39. How can you find the correlation of two sequences using convolution
40. what is the constraint for location of poles in a causal stable system
41. What is pole zero cancellations and is it advisable to follow this approach
42. What is an all pass filter. Give some examples. what are they used for
43. What is the advantage of a Direct form II FIR over form I realizations?
44. What does QM.N representation mean for signed and unsigned integers.
45. what is the value of decimal 2398 when looked in Q3.12 and similarly 1.5 in Q3.12 corresponds to which decimal value. it is assumed these are unsigned numbers
46. what is the value of decimal 2398 when looked in Q3.12 and similarly -1.5 in Q3.12 corresponds to which decimal value. Assume the numbers are signed
47. What is the resulting Q format obtained when two numbers of QM1.N1 and QM2.N2 are added, substracted, multiplied and divided. Answer for both signed and unsigned numbers and there should not be any truncation of the results.
48. If the result is to be truncated to QM1.N1 In the above question, how do you truncate in all the above operations for singed and unsigned numbers.
49. What is the minimum and maximum value represented by signed and unsigned numbers in QM.N format. Also, what is their precision.
50. What is the minimum and maximum value represented by signed and unsigned numbers in QM.N format. Also, what is their precision.
60. What is paley weiner theorem
2. How to find the convolution of two sequences using DFT operation
3. What is the difference between DFT and DTFT. What is the relation between them.
4. What is the need for DFT when DTFT is already present
5. How to find the dft of two N point real sequences using a N point DFT
6. What is the reduction in computational complexity by implementing DFT as FFT
7. If you have an N point dft calculator and two N/2 sequences. How can you find the dft of the sequences
8. How to find the dft of a 2N point real sequence with a N point DFT hardware
9. What are overlap save and overlap add methods and what are they used for
10. If we limit a signal to L samples, what is the maximum difference between frequencies that will be seen as same frequency.
11. Describe the process of analog to digital conversion and the effect of quantization noise on resulting SNR (where noise is quantization noise)
12. What is the application of Cross correlation and Auto Correlation
13. If you have a signal x[n] and it is upsampled by N and then downsampled by M. what is the condition for not losing any information in x[n]
14. How do u reduce spectral leakage
15. What is the speciality about minimum phase filter
16. why IIR filters doesn't have Linear phase
17. Can IIR filters be Linear phase? how to make it linear Phase?
18. Describe the process of decimation. what is the cutoff frequency of the filter to be used and why is it required
19. Describe the process of Interpolation. what is the cutoff frequency of the filter to be used and why is it required
20. What is the simplest high pass filter. Write the equation
21. What is the difference between equiripple filter and FIR filter
22. What’s differences b/w butter worth, chebyshev, elliptical filter and advantages/disadvantages of each
23. why we use DCT extensively in compression
24. Why after DCT we use a zig zag manner for run length coding
25. What is the basic difference between FIR and IIR filters
26. What are limit cycle oscillations
27. Is the system y = ax + n linear
28. If you have 16 bit multiplier hardware, how do you find 32 bit multiplication
29. If there is a discontinuity in the signal amplitude in time, why will there be more spectral components
30. If we have a 10 bit input sample and an FIR filter with 15 taps each of 6 bits, what is the minimum number of bits required to store the output
31. What is the importance of linear phase of a filter and it’s relation to group delay
32. How does the location of poles and their order affect the stability of a system
33. What is a continuous time signal, discrete signal and digital signal
34. Describe discrete time sinusoid signal. What is the condition for the signal to be periodic
35. How are signals classified into energy and power signals
36. What are causal systems
37. When is a system said to be stable
38. What is the condition for stability of linear time invariant systems
39. How can you find the correlation of two sequences using convolution
40. what is the constraint for location of poles in a causal stable system
41. What is pole zero cancellations and is it advisable to follow this approach
42. What is an all pass filter. Give some examples. what are they used for
43. What is the advantage of a Direct form II FIR over form I realizations?
44. What does QM.N representation mean for signed and unsigned integers.
45. what is the value of decimal 2398 when looked in Q3.12 and similarly 1.5 in Q3.12 corresponds to which decimal value. it is assumed these are unsigned numbers
46. what is the value of decimal 2398 when looked in Q3.12 and similarly -1.5 in Q3.12 corresponds to which decimal value. Assume the numbers are signed
47. What is the resulting Q format obtained when two numbers of QM1.N1 and QM2.N2 are added, substracted, multiplied and divided. Answer for both signed and unsigned numbers and there should not be any truncation of the results.
48. If the result is to be truncated to QM1.N1 In the above question, how do you truncate in all the above operations for singed and unsigned numbers.
49. What is the minimum and maximum value represented by signed and unsigned numbers in QM.N format. Also, what is their precision.
50. What is the minimum and maximum value represented by signed and unsigned numbers in QM.N format. Also, what is their precision.
60. What is paley weiner theorem
The questions and answers are not corresponding.
ReplyDeleteAgree, and its annoying.
ReplyDelete5. How to find the dft of two N point real sequences using a N point DFT
ReplyDeleteAnswer: Make the sequence in two group odd and even, pass the odd sequence with one N point DFT and even component with another N point DFT. then combine the output of one FFT group with other with weight factor = exp(-2*pi*n/2N) as we used to do in last step of fft butterfly diagram. Please refer fft diagram for it
10. If we limit a signal to L samples, what is the maximum difference between frequencies that will be seen as same frequency.
ReplyDeleteAnswer: The question state that, if we only take L samples from a signal ( windowing it to L samples) then what is the max frequency till which we can't distinguish between two signal. As we know for taking only L samples we need to multiple the sampled input with some window of length L ( which can make other component as zero). multiplication in time domain is convolution in freq domain, which is convolution with sinc pulse (fourier transform of window function) with actual spectrum. The zero crossing of sinc pulse will occur at 2*pi/L so the max seperation with which cant distinguish is 1/LT. Please refer Digital Signal Processing: Principles, Algorithms, And Applications, By John G. Proakis section 7.4
If you are preparing for physical layer interviews, you can checkout Tele Interview
ReplyDeletehttps://goo.gl/2kIs1Q
https://www.facebook.com/Tele-Interview-224846541330838/
5- If g(n) and h(n) are two real sequence, we should calculate DFT of signal x(n) = g(n) + jh(n).
ReplyDeleteG[k]= 0.5 * (X[k] + X[-k]*)
H[k] = -0.5j * (X[k] - X[-k]*).
4- DTFT of a periodic signal diverge. For periodic signal, DFT has been defined.
ReplyDelete6- The complextiy computation for DFT is O(N^2). The complexity computation of FFT is O(NlogN).
ReplyDelete8- Assume v[n] has 2N samples, we consider h[n]=v[2n] and g[n]=h[2n+1].
ReplyDelete9. What are overlap save and overlap add methods and what are they used for
ReplyDeleteIf we have a large sequence data x[n] and want to calculate y[n] = x[n] * h[n]('*' is convolution), it will be time consuming. There is two method for easier calculation y[n]:
a) overlap-add method
b) overlap-save method
these methods divide data x[n] to different parts of length L.
overlap-add method:
1. Break the input signal x(n) into non-overlapping blocks x_m(n) of length L.
2. Zero pad h(n) to be of length N = L + M - 1.(M is the length of h(n))
3. Take N-DFT of h(n) to give H(k), k = 0; 1; : : : ;N - 1.
4. For each block m:
4.1 Zero pad x_m(n) to be of length N = L + M - 1.
4.2 Take N-DFT of x_m(n) to give Xm(k), k = 0; 1; ... ;N - 1.
4.3 Multiply: Y_m(k) = X_m(k) * H(k), k = 0; 1; ... ;N - 1.
4.4 Take N-IDFT of Y_m(k) to give y_m(n), n = 0; 1; ... ;N - 1.
5. Form y(n) by overlapping the last M - 1 samples of ym(n) with the rest
M - 1 samples of ym+1(n) and adding the result.
overlap-save method
1. Take N-DFT of xm(n) to give Xm(k), k = 0; 1; ... ;N - 1.
2. Take N-DFT of h(n) to give H(k), k = 0; 1; ... ;N - 1.
3. Multiply: Ym(k) = Xm(k) * H(k), k = 0; 1; ... ;N - 1.
4. Take N-IDFT of Ym(k) to give yC;m(n), n = 0; 1;... ;N - 1.
x_m[n] in overlap-save method will be padded with M-1 samples from x_m-1[n].
12-What is the application of Cross correlation and Auto Correlation?
ReplyDeleteThey can be used for signal detection in radar, pattern detection. Cross correlation measures the similarity between two signals.
13. If you have a signal x[n] and it is upsampled by N and then downsampled by M. what is the condition for not losing any information in x[n]
ReplyDeleteUpsampling:
x_z[n]=\sum_{k=-inf}^{inf}x[n]\delta(n-kM)
X_z(jw) = X(jWM)
After adding zero between samples, we pass the signal X_z(jw) through a lowpass filter.
Downsampling
1- Low pass signal to omit undesired part.
2- x_z(n) = x(Mn)
14. How do u reduce spectral leakage?
ReplyDeleteIf we have a signal and sample it, in frequency domain, spectral leakage happens because sampling a signal in a finite time is such as multiplying signal by a rectangular window in time.
To overcome this problem, we apply smooth window that smootly goes to zero.(Hanning window)
We multiply the sampled signal by smooting window.
Delete35. How are signals classified into energy and power signals
ReplyDeleteIf a signal has bounded power said to be power signal. If its energy is bounded said energy signal.
36. What are causal systems
The system that its output depends to the past and current input not future input.
37. When is a system said to be stable
If for a bounded input, output is bounded
38. What is the condition for stability of linear time invariant systems?
the integral of amplitude of channel be bounded
What is the speciality about minimum phase filter?
ReplyDeleteIt has minimum group delay.
16.why IIR filters doesn't have Linear phase?
ReplyDeleteIn order to be linear phase, H(z)=H(1/z) which implies have poles inside and outside unit circle and it makes the system unstable.