Discussion Forums

Gauss-Markov mode filtering

No replies
sigblips
sigblips's picture
Offline
Joined: 2010-04-20
Posts: 732

Jill Tarter asked me to post these notes from a lecture she attended:

"This past Wednesday I attended a lecture at WHOI (Woods Hole Oceanographic Institute) about passive echo location of Right and Sei whales in Cape Cod harbor using triangulation from 3 vertical and horizontal hydrophone bouys. One thing I found interesting is that the whales emit a particular call the researchers call a ‘gunshot’ (short pulse covering multiple frequencies) which then exhibits modal dispersion as it propagates through the salt water. [Think of an impulse response function sampling the interstellar medium.] I wondered whether the animals can actually sense this dispersion and use it to estimate the distance of the sender --- who knows? Since dispersion is very sensitive function of water temperature and salinity, this might not make any evolutionary sense. BUT the thing that caught my attention was during the question period when the researcher mentioned use of Gauss Markov mode filtering to separate signal from noise+signal without requiring a model of the signal. Seems to me this has the potential for becoming a good anomaly detector for SETI. Does anyone know anything about this technique? How compute-intensive is it? Does it parallelize well?"

I didn't know much about this technique so I did some reseach. Here are some reading links to get you started learning more about the Gauss-Markov model:

http://www.cs.cmu.edu/~jch1/research/gaussmarkov/gaussmarkov.pdf
http://en.wikipedia.org/wiki/Gauss-Markov_process
http://www.stanford.edu/class/ee363/lectures/kf.pdf

The next steps would be:

  1. Figure out how useful and applicable it is to SETI
  2. Prototype it
  3. Test it with setiData to measure how well it works
  4. Optimize and implement it in SonATA

This would be a great project for someone with a bit of math and science background to work on.