1
$\begingroup$

I'm a bit confused about the way echo canellation over a voice telephone network works like in the following figure enter image description here

My issue is with the fact that the point of the adaptive filter and LMS algorithim is to drive the error e(t) to zero correct? Then surely, Speaker B would hear nothing as the algorithim converges.

$\endgroup$
  • 1
    $\begingroup$ think about what your $e(t)$ represents. $\endgroup$ – Marcus Müller Dec 5 '18 at 18:32
  • $\begingroup$ @MarcusMüller It repersents the unechoed, original speech of speaker A that I want speaker B to hear right? $\endgroup$ – AlfroJang80 Dec 10 '18 at 2:46
1
$\begingroup$

A usual arrangement of echo cancelers in a full-duplex system is shown in the following figure (from Digital Communication by Lee and Messerschmitt):

enter image description here

The echo cancelers try to remove any residual transmitted signal in the receiver of the same end. Since the input to the echo canceler is the transmitted signal, it can only remove signal components in the received signal that are correlated with the transmitted signal. It cannot remove any other signal components, such as the signal that is transmitted from the far end (which is the desired received signal). Consequently, the error signal can never become zero, and the desired signal (transmitted from the other end) will not be influenced by the adaptive filter.

$\endgroup$
  • $\begingroup$ Ah. That makes so much more sense. Thank you so much. $\endgroup$ – AlfroJang80 Dec 10 '18 at 2:48

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.