I just want to ask, whenever you reduce a bandwidth of signal by filtering, I wonder why do you need to improve the number of bits used to represent that signal. The number of bit growth is $$\frac{1}{2}\log_2\left(\frac{BW_{in}}{BW_{out}}\right)$$

Is it to represent to small noise due to filtering. Thanks!

  • $\begingroup$ Is there a specific paper or book you're using? Where is this formula from? $\endgroup$
    – Phonon
    Oct 24 '13 at 1:50
  • $\begingroup$ Perhaps my knowledge is limited, but I never knew that bandwidth and bits used for representing a signal are related. Further, if noise is a deciding factor, I don't think we need to increase the number of bits, since noise in general may be white or have more high frequency content, which means that filtering only reduces the noise. $\endgroup$
    – Vishwanath
    Oct 24 '13 at 4:09
  • $\begingroup$ It's from my mentor, he said it has something to do with the signal to noise ratio improvement since you filter out noise, hence you need to add a certain number of bits to achieve that Signal to Noise Ration. I really can't understand it quite well. $\endgroup$
    – Cordic
    Oct 25 '13 at 8:04

Consider a two-sample moving average of one-bit data with values {0,1}. The bandwidth is reduced by a factor of two. The possible outputs are {0,0.5,1}. You have gained half a bit, as predicted by the formula.


  • $\begingroup$ How come you reduce the bandwidth by two, you have those 3 possible outputs --> (0, 0.5, 1)? $\endgroup$
    – Cordic
    Oct 25 '13 at 8:08
  • $\begingroup$ (0 + 0) / 2 = 0, (1 + 0) / 2 = 0.5, (1 + 1) / 2 = 1 $\endgroup$
    – John
    Oct 30 '13 at 0:51

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.