At 11:41 AM 7/11/2007, Bob McGwier wrote:
>Here is the grand experiment we are going to try when we finally get to
>calm down from the Flex5000 rush to finish.   I have done enough
>matlab/octave experimentation to expect it to work so long as we define
>what it is we mean by work.
>
>Suppose you are listening to weak signal A.   Strong signal B comes on
>and its main power is well out of your passband but the splatter or
>sidebands are in your passband and harms your ability to hear signal
>A.    The mathematical idea is that the portion of the signal B that is
>in your passband is correlated with the "main" signal that is out of
>your passband.  It is strongly correlated and we should be able to
>derive a "filter" that will predict a version of the signal that is IN
>your passband.  The property of the signal we will optimize on is to
>reduce the energy of the interference to the best of our ability.   If
>we get 20 dB reduction of the inband interference from an "out of band"
>interferer, I will consider it a victory.   Much more than that and I
>will consider it to be a major league success.



AKA an adaptive signal canceller.  You adapt to find the strongest 
signal (in time/frequency, as with radar STAP), subtract it from the 
input. Then find the next strongest signal, subtract it, etc.  MUSIC, 
ESPRIT, etc. all implement various forms of this.

With something like PSK31 as the interfering signal (or, possibly CW, 
and even more remotely possible, ssb voice), you've actually got some 
side information as to the structure of the interfering signal, which 
reduces the space over which the algorithm has to "work".  It's 
basically a signal estimation problem, and there's a lot of theory 
out there to draw from, both in radio and acoustics.

Early significant similar applications were cancelling 60 Hz 
artifacts in EKG and EEGs by guys like Widrow in the 1960s. There was 
a good review article in IEEE Proceedings in the late 70s, early80s 
that talked about this. The papers by Schmidt describe how MUSIC 
works quite nicely.  Lately, there's been a lot of work using 
multiple antennas, so you get spectral, spatial, and temporal information.



Jim, W6RMK



_______________________________________________
FlexRadio mailing list
FlexRadio@flex-radio.biz
http://mail.flex-radio.biz/mailman/listinfo/flexradio_flex-radio.biz
Archive Link: http://www.mail-archive.com/flexradio%40flex-radio.biz/
FlexRadio Knowledge Base: http://kb.flex-radio.com/
FlexRadio Homepage: http://www.flex-radio.com/

Reply via email to