hydrophone array processing on a beowulf

[Morton, Scott]

The petroleum industry is rapidly adopting beowulfs for the multichannel
acoustic/elastic signal processing involved in searching for oil and gas. We
are "looking for largish things", specifically deposits of hydrocarbons.

While we don't generally analyze our results in real time, the data volumes
and computations are quite large. The number of channels in a typical marine
survey is in thousands and the number of experiments in the hundreds of
thousands, yielding TBs of data.  The computation to produce an image of the
subsurface can be measured in cpu-decades. Fortunately most of the
computation involved is either trivially parallelizable or is
computationally intensive enough that communication speeds are not a serious
issue, so simple, large-scale beowulfs are a natural and cost-effective
solution for us.

I gave a talk at SC99 which you could use as a short intro to seismic
imaging and our use of beowulfs to perform these computations. If you're
interested, see www.sc99.org/proceedings/papers/morton.pdf.

I konw the Navy does this kind of processing also, though since their
systems are probably all on-board ships and subs, they probably use
ruggedized systems instead of standard beowulfs.

Scott Morton

-----Original Message-----
From: Jim Lux [mailto:jimlux at jpl.nasa.gov]
Sent: Sunday, February 25, 2001 11:08 AM
To: beowulf at beowulf.org
Subject: hydrophone array processing on a beowulf


Inspired by the recent Nova on Nessie, we had a discussion about the
feasibility of setting up a hydrophone array to constantly monitor looking
for largish things, prompting the question whether there are any open
literature references to the algorithms used, and even better, has anyone
done this sort of multichannel acoustic signal processing on a beowulf.



_______________________________________________
Beowulf mailing list, Beowulf at beowulf.org
To change your subscription (digest mode or unsubscribe) visit
http://www.beowulf.org/mailman/listinfo/beowulf