In many real applications, practical constraints do not allow long time sequences to be processed. However, significant information can be gained from a windowed section of the sequence. In this problem, you will look at computing the Fourier transform of an infinite-duration signal x[n], given only a block of 256 samples in the range 0 ≤ n ≤ 255. You decide to use a 256-point DFT to estimate the transform by defining the signal
and computing the 256-point DFT of
(a) Suppose the signal x[n] came from sampling a continuous-time signal xc(t) with sampling frequency fs = 20 kHz; i.e.,
x[n] = xc(nTs),
1/Ts = 20 kHz.
Assume that xc(t) is bandlimited to 10 kHz. If the DFT of 0, 1, . . . , 255, what are the continuous-time frequencies corresponding to the DFT indices k = 32 and k = 231? Be sure to express your answers in Hertz.
(b) Express the DTFT of in terms of the DTFT of x[n] and the DTFT of a 256-point rectangular window wR [n]. Use the notation X (ejω) and WR(ejω) to represent the DTFTs of x[n] and wR [n], respectively
(c) Suppose you try an averaging technique to estimate the transform for k = 32:
Averaging in this manner is equivalent to multiplying the signal by a new window wavg[n] before computing the DFT. Show that Wavg(ejω) must satisfy
where L = 256.
(d) Show that the DTFT of this new window can be written in terms of WR(ejω) and two shifted versions of WR(ejω).
(e) Derive a simple formula for wavg[n], and sketch the window for α = −0.5 and 0 ≤ n ≤ 255.
We need at least 10 more requests to produce the solution.
0 / 10 have requested this problem solution
The more requests, the faster the answer.