a)Draw the frequency spectrum of an analog signal after sampling when the sampling frequency is 1.70 MHz and the source signal has a frequency range (i) 100 – 500 kHz, (ii) 100 – 700 kHz, and (iii) 100 – 900 kHz. (For sketching purposes, you may assume that the source signal strength is uniform over the signal bandwidth.) If we use a lowpass filter of bandwidth 900 kHz before sampling, what happens in each case when we attempt to reconstruct the original signal from the sampled signal? In each case, if there are unwanted frequency components in the reconstructed signal, what is the primary cause, and can we remove these components before sampling?

 

b)In sub-band coding, the frequency spectrum of the media signal is subdivided into smaller bands and then each sub-band is coded separately (in fact, different source coding techniques may be applied in each sub-band). Sub-band coding is to be used by a super-wideband audio codec on a music stream with linear PCM applied separately in each sub-band. Assume the total audio range is from 0 Hz to 21 kHz and 7 sub-bands of equal size are used. 8 bits are then assigned to samples from the lowest 2 sub-bands, 6 bits to samples from each of the next 2 lowest sub-bands, and 4 bits to samples from the remaining sub-bands.

Calculate the overall data rate if the sampling rate for each sub-band is 5% higher than the Nyquist rate.

If linear quantization is used in each of the sub-bands, how many quantization intervals are needed in each case?