Ganesh,
There is a lot of theory behind that, but practically the leaky LMS has
a few advantages - it forces the filter taps to keep learning in all of
the frequency bands constantly. So if the input signal doesn't excite
all of the frequency components (zero-value eigenvalues of the
autocorrelation matrix), the filter taps are not going to converge, the
filter is not going to converge to a unique solution... That's why the
leakage is used to keep the taps changing in all of the frequency
components, and it's changing them towards zero - preventing the noise
from building up. I've heard that it also helps with the finite
precision math.
Vlad
-----Original Message-----
From: ganesa1234 [mailto:]
Sent: Tuesday, September 21, 2004 11:02 PM
To:
Subject: [echocancel] Leaky NLMS
Dear Members
I am Ganesh, working on Echo Cancellation for VoIP networks.
While doing my simulations with a few different algorithms like LMS and
NLMS, i didnt get a very satisfactory result.
But when i used Leaky NLMS it works perfectly. Does anyone know
why it works very well for Leaky NLMS? If you'll know of any useful
links also, please let me know.
Many Thanks
Hope to hear from you'll soon.
Regards
Ganesh
------------------------ Yahoo! Groups Sponsor --------------------~-->
Make a clean sweep of pop-up ads. Yahoo! Companion Toolbar.
Now with Pop-Up Blocker. Get it for free!
http://us.click.yahoo.com/L5YrjA/eSIIAA/yQLSAA/sVPplB/TM
--------------------------------~->
_____________________________________
Note: If you do a simple "reply" with your email client, only the
author
of this message will receive your answer. You need to do a "reply
all"
if you want your answer to be distributed to the entire group.
_____________________________________
About this discussion group:
To Join:
To Post:
To Leave:
Archives: http://www.yahoogroups.com/group/echocancel
Other DSP-Related Groups: http://www.dsprelated.com
Yahoo! Groups Links
|