Download as docx, pdf, or txt
Download as docx, pdf, or txt
You are on page 1of 2

EE684A: Fiber-Optic Communications

Due on 25/08/2017

Assignment #2

1. Let 𝑥(𝑛) be a 8-point sequence given by

0 𝑛<0
𝜋
𝑥(𝑛) = {cos ( 𝑛) 0 ≤ 𝑛 ≤ 7
4
0 𝑛≥8

and 𝑦(𝑛) be a 8-point sequence given by

0 𝑛<0
𝜋
𝑦(𝑛) = {cos ( 𝑛) 0 ≤ 𝑛 ≤ 7
3
0 𝑛≥8

a) Compute the 8-point DFT of 𝑥(𝑛) and sketch a plot of the magnitude of the DFT.
b) Repeat (a) for the sequence 𝑦(𝑛).
c) Compare your answers from parts (a) and (b). Explain the differences, if any.

2. Consider the real-valued signal 𝑠(𝑡) whose Fourier transform is given below. Sketch the DTFT of
the sequence resulting from sampling 𝑠(𝑡) at 8 kSa/s. Be sure to properly label your axes.

3. The even part of a real sequence 𝑥[𝑛] is defined by


𝑥[𝑛]+𝑥[−𝑛]
𝑥𝑒 [𝑛] = 2
.
Suppose that 𝑥[𝑛] is a real finite length sequence defined such that 𝑥[𝑛] = 0 for 𝑛 < 0 and 𝑛 ≥ 𝑁.
Let 𝑋[𝑘] denote the N-point DFT of 𝑥[𝑛].

a) Is Re{𝑋[𝑘]} the DFT of 𝑥𝑒 [𝑛] ?


b) What is the inverse DFT of Re{𝑋[𝑘]} in terms of 𝑥[𝑛]?

4. This problem must be solved using MATLAB.

a) Construct the 128-point sequence 𝑥[𝑛] to be a cosine wave of 0.125 cycles per sample plus 0.5
times a sine wave of 0.25 cycles per sample. Using MATLAB, compute the DFT of the sequence 𝑥[𝑛].
Plot the magnitude of the DFT.

b) Pad 896 zeros at the end of the 128-point sequence and use MATLAB to compute the 1024-point
DFT of the zero-padded sequence. Does the 1024-point DFT look the same as the 128-point DFT of
part (a)? If not, how is it different?
c) Now find the 1024-point DFT of a pulse sequence with width 128 (𝑥[𝑛] = for 𝑛 =
0,1,2, … . . ,127 and 0 otherwise). How would this DFT explain the differences between (a) and (b)?
5. Consider the continuous-time signal
𝑥(𝑡) = cos(10𝜋𝑡) + sin(3𝜋𝑡 + 𝜋/2).
Let T be the sampling interval. We uniformly sample 𝑥(𝑡) to obtain the sequence 𝑥[𝑛𝑇].

a) What is the largest value of T that we can use to sample the signal and still achieve perfect
reconstruction?
b) Suppose we sample with T=1/30. Illustrate the CTFT of 𝑥(𝑡), CTFT of sampled signal, and DTFT of
sampled sequence in the interval from -1/2T to 1/2T. Label correctly the axes.
c) Repeat (b) for T=1/20.

6. Let 𝑥[𝑛] be a random process generated by


𝑥[𝑛] = 𝑎𝑥[𝑛 − 1] + 𝑤[𝑛],
where 𝑛 ≥ 0, 𝑥[−1] = 0, and 𝑤[𝑛] is an IID (𝜇𝑤 , 𝜎𝑤2 ) process.

a) Show that the process 𝑥[𝑛] can be expressed as


𝑛

𝑥[𝑛] = ∑ 𝑎𝑛−𝑖 𝑤[𝑖].


𝑖=0
b) Show that the mean of the process 𝑥[𝑛] for a=1 is (𝑛 + 1)𝜇𝑤 .
c) Show that for 𝑎 ≠ 1, the mean is
𝜇𝑤 𝑎𝑛 (1 − 𝑎−(𝑛+1) )
.
1 − 𝑎−1
d) What can you say about the mean when |𝑎| < 1 in the limit of 𝑛 → ∞?

7. Use MATLAB for this problem. Consider a discrete time IID Gaussian WSS process with zero mean
and unit variance, denoted by 𝑤[𝑛]. Let 𝑥[𝑛] = 𝑤[𝑛] + 0.5𝑤[𝑛 − 1] + 0.25𝑤[𝑛 − 2]. Generate a
realization of length 256 of 𝑥[𝑛].

a) Plot the time averaged mean, 𝑚


̂𝑥 defined as
𝑁−1
1
𝑚
̂𝑥 = ∑ 𝑥[𝑛],
𝑁
𝑛=0
for 𝑁 = 5, 10, 20, 50, 75, 100, and 256. Explain your results.
̂𝑥 [𝑘] defined as
b) Plot the time averaged autocorrelation, 𝑅
𝑁−1−𝑘
1
̂𝑥 [𝑘] =
𝑅 ∑ 𝑥[𝑛]𝑥[𝑛 + 𝑘],
𝑁−𝑘
𝑛=0
for 𝑘 = 0, 1, and 2 and same values of 𝑁 as in part (a). Compare your results with analytically
computed result.

8. Using MATLAB, generate at least three realizations of discrete laser phase noise process modelled
as a Weiner process discussed in class. Assume that the zero mean IID 𝑤[𝑛] in the Weiner process
have zero mean and variance 𝜋. Plot them on the same graph and comment on your results.

You might also like