Often in estimation of the input signal to an unknown channel, the bit error rate is plotted as the performance measure. The curve is compared with an ideal curve labelled as 'Perfect channel'. For example in this paper, http://www.iaeng.org/publication/WCECS2015/WCECS2015_pp668-673.pdf although the bit error rate curve is not present, in the introduction the Authors do mention the term perfect channel. One such method of estimation is the Least Squares.

What is the meaning of a perfect channel? Does a perfect channel mean that there is no noise or does it mean that the pilot symbols are used to estimate the channel?


1 Answer 1


It's not the channel that is perfect; it is the estimation. So, "perfect channel estimation" means that the receiver knows the fading coefficients perfectly. This is of course only true in a simulation; any actual system performs imperfect estimation and the estimated channel coefficients are only approximately equal to the actual coefficients.

  • 1
    $\begingroup$ All except the simplest communications systems need channel estimation; in particular, any system where the signal bandwidth is large enough that the channel introduces intersymbol interference. This includes both wireline (e.g. data communications over the telephone line) and wireless systems. I don't know a lot about controls, but system identification is very similar theoretically; the difference is that you have a general system (for example, a loaded motor) instead of a comm channel. $\endgroup$
    – MBaz
    Commented Dec 9, 2017 at 21:10
  • $\begingroup$ Here's an example of equalization for data comm over the telephone local loop: ieeexplore.ieee.org/document/841833 $\endgroup$
    – MBaz
    Commented Dec 9, 2017 at 21:13

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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