Download as pdf or txt
Download as pdf or txt
You are on page 1of 3

CNAM/MRI02 Februray 17nd 2023

ELE102: Digital Signal Processing


Final Exam

All papers documents authorized

Duration: 3h00

Exercise n°1:
We consider the following analog signal

𝑥(𝑡) = 1 + 2cos(2𝜋𝑓1 𝑡) + 2cos⁡(2𝜋𝑓2 𝑡)

where⁡𝑓1 = 10𝑘𝐻𝑧 et 𝑓2 = 30𝑘𝐻𝑧

1. Compute the period of the signal 𝑥(𝑡) and its average power.
2. Compute and plot the modulus of the Fourier transform |𝑋(𝑓)| of the signal 𝑥(𝑡).
3. We sample this signal at a sampling frequency⁡Fs = 50⁡kHz,⁡to get a signal xs (t). Give the analytical
expression of the sampled signal spectrum 𝑋𝑠 (𝑓).
|𝑋𝑠 (𝑓)|
4. Plot the modulus of the normalized spectrum , over the frequency band [−60⁡kHz, 60⁡kHz]
𝐹𝑠
and analyze the obtained result.
5. Is-it possible to recover the original signal after filtering by a perfect reconstruction filter: a low pass
𝐹
over band [0, 2𝑠 ]. Justify your answer.
6. We modify the sampling frequency 𝐹𝑠 to 𝐹𝑠′ . The new discrete time signal is now given by:

𝑥 ′ (𝑛) = 1 + 2 cos(𝜋𝑛/5) + 2cos⁡(3𝜋𝑛/5)

Compute the new sampling frequency 𝐹𝑠′ and explain if the frequency is compliant with the Shannon
theorem.
7. The signal 𝑥 ′ (𝑛) passes through an analog to digital 16-bit converter (ADC), which limits (clips) the
sample amplitudes below -5V and above + 5V to -5V and + 5V, respectively. Compute the crest factor
of this signal.
8. Compute the signal-to-noise ratio at the ADC output.
9. Compute the quantization step size of the ADC.
10. Compute the power of the quantization noise.
11. Based on results of questions 1) and 10) compute the signal-to-noise ratio at the ADC output and
compare it to the result of question 8.
12. What is the memory space, in bytes, needed to store 100 signal periods?

Exercise n°2:
A discrete time domain signal 𝑥(𝑛) is filtered with a filter of transfer function equal to 𝐻(𝑧).
The signal at the output of this filter is in turn filtered by two filters 𝑆1 (𝑧) and 𝑆2 (𝑧), to give two signals
𝑑(𝑛) and 𝑔(𝑛). The figure below shows the block diagram of the considered system.

Digital Signal Processing ELE102 1


CNAM/MRI02 Februray 17nd 2023

• Study of the filter H(z)


The poles and zeros of the filter 𝐻(𝑧) are placed in the complex plane according to the
following figure:

1. Compute the transfer function, 𝐻(𝑧), of this filter.


2. Is this filter FIR or IIR? Justify your answer.
3. Discuss the stability of this filter.
4. Write the signal 𝑦(𝑛), as function of the input signal 𝑥(𝑛).
5. Compute the frequency response, 𝐻(𝑓), of this filter.
6. Compute the filter gain at 𝑓 = ±𝐹𝑠 /4. (𝐹𝑠 is the sampling frequency).

• Study of the filter S1(z)


The signal at the output of the filter 𝑆1 (𝑧) satisfies the following difference equation:

𝑑(𝑛) = 0.5(𝑦(𝑛) − 𝑦(𝑛 − 1))

7. Compute the impulse response, 𝑠1 (𝑛), of this filter.


8. Compute the transfer function, 𝑆1 (𝑧), of this filter.
9. Compute the frequency response, 𝑆1 (𝑓), of this filter.
10. What’s the shape of the filter response over the frequency band [0⁡𝐹𝑠 /2]?
11. Compute the phase response of this filter 𝜑𝑠1 (𝑓) and discuss its behavior with respect to
frequency?

• Study of the filter S2(z)


The implementation scheme of this filter is given by the following figure.

Digital Signal Processing ELE102 2


CNAM/MRI02 Februray 17nd 2023

12. Write the signal 𝑔(𝑛), at the output of this filter as function of the input signal 𝑦(𝑛).
13. Compute the impulse response, 𝑠2 (𝑛), of this filter.
14. Compute the transfer function, 𝑆2 (𝑧), of this filter.
15. Compute the frequency response, 𝑆2 (𝑓), of this filter.
16. What’s the shape of the filter amplitude response over the frequency band [0⁡𝐹𝑠 /2]?

1.1. Filtering of a signal x(n)


We consider a discrete time domain signal 𝑥(𝑛), at the input of the entire system given by:

𝑥(𝑛) = 1 + cos(2𝜋𝑓1 𝑛/𝐹𝑒 ) + cos⁡(2𝜋𝑓2 𝑛/𝐹𝑒 )

where⁡𝑓1 = 0.25⁡𝐹𝑠 , 𝑓2 = 0.45⁡𝐹𝑠 and 𝐹𝑠 = 100𝐻𝑧.

17. Compute the transfer function of the system between 𝑥(𝑛) and 𝑑(𝑛).
18. Write the difference equation between 𝑥(𝑛) and 𝑑(𝑛).
19. Compute the transfer function of the system between 𝑥(𝑛) and 𝑔(𝑛).
20. Write the difference equation between 𝑥(𝑛) and 𝑔(𝑛).
21. Compute the modulus of the first component of the DFT of the signal 𝑔(𝑛) : |𝐺(0)|, computed
over 𝑁 = 100 points.

Exercise n°3
π⁄
We consider a second order FIR filter, with complex et conjugate zeros given by: z1 = ej 4 et
∗ −jπ⁄4
z1 = e . This filter processes a normalized and zero mean white noise b(n), (E[b(n)] = 0) and
(σ = E[b(n)2 ] = 1). We note 𝑥(𝑛) the signal at the output of the filter.
2

1. Compute the coefficients of the filter impulse response: h(𝑛).


2. Compute the transfer function of this filter: 𝐻(𝑧) = 𝑍𝑇(h(𝑛)).
3. Compute the frequency response of this filter: 𝐻(𝑓) = 𝐹𝑇(h(𝑛)).
4. Compute the response of this filter at zero frequency.
5. Which normalized frequency is totally suppressed by this filter?
6. Write 𝑥(𝑛) according to the input signal.
7. Compute the autocorrelation coefficients 𝑟𝑥𝑥 (𝑝) of 𝑥(𝑛) for 𝑝 = 0, 𝑝 = 1, 𝑝 = 2.
8. Give the general form of 𝑟𝑥𝑥 (𝑝).
9. Is the signal 𝑥(𝑛)n) white? justify your answer.
10. Compute the power spectral density 𝑆𝑥 (𝑓): of the signal 𝑥(𝑛).
11. Demonstrate that 𝑆𝑥 (𝑓) = |𝐻(𝑓)|2.

Good luck…

Digital Signal Processing ELE102 3

You might also like