[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: SETI like tass project
I seem to recall a study by NASA that concluded that the highest possible
network bandwidth is achieved by shipping a box load of CDs via DHL or
FedEx. Not quite the SETI model, however as the data reduction is not
particularly time dependant good old "snail mail" would work, provided the
owners of the instruments are prepared to make the additional CDs.
I have an old PC I could resurrect if CPU speed is not critical, however
like Tom I am not particularly keen on learning LINUX from scratch. Would it
be possible to distribute an installation CD that would set up the operating
system and all the necessary software for the pipeline?
----- Original Message -----
From: Chris Albertson <chrisalbertson90278@yahoo.com>
To: <tass@listserv.wwa.com>
Sent: Wednesday, January 30, 2002 9:54 AM
Subject: Re: SETI like tass project
>
> Tom,
>
> With SETI there is a small "work packet" that can be transmitted in
> a minute or so even using a modem. On a typical PC each work
> packet takes a tens of minutes to process. So with SETI the ratio
> of data transport to computation is very good.
>
> With TASS the smallest "work packet" is one image which is
> 2K x 2k x 16 bits = 8M. These typically compress _without_ loss
> to about 6.5MB per image.
> Still, It is reasonable, even with a modem to download one 6.5MB image
> with many people doing the downloads an entire nights worth of
> data could be reduced in under one hour. BUT, (there is always
> a "but") these images would have to be placed on a server that
> has _very_ high bandwidth connection to the Internet. The server
> would need enough bandwidth to push one night's worth of the data
> out each day. Worse, it needs to be twice that fast at least,
> if users aren't to see it as being uselessly slow and give up
> on the downloads. Still worse, the SETI-like clients will
> likely NOT request data at a uniform rate but it will peak over
> each 24 hour period. We are talking "multi-megabits/second".
> DSL or Cable modems just would NOT cut it for the server. This is
> possible but the cost of such bandwidth is about $1k/month.
> We would flood even some university's Internet connections
>
> There is also the possibility of using multiple image servers
> but with multiple TASS camera sites this would be a given.
> At any rate the "one night's data per day" requirement is a
> demanding (and expensive) constraint.
>
> Another option would be to install multiple computers at the
> camera sites. I can envision a "black box" with no keyboard
> or CRT, just an Ethernet connection and power cord.(not even a
> CDROM or floppy.) These "black boxes" connect to the local
> Ethernet and request data, process it and send the results back.
> They should be designed so that one could connect any number of
> these black boxes to the same local Ethernet so the processing
> rate would scale. User training is minimized as there is only
> one user control (the power switch.)
>
> This would be the SETI model but using 100Mb/sec communication
> media. I think with current technology moving computers is
> cheaper then moving data.
>
>
>
> --- Tom Droege <tdroege2@earthlink.net> wrote:
> > I have accumulated a lot of data. About 375 CD's at this point. I
> > figure
> > another 1000 before the year is out.
> >
> > I would again like to bring up the possibility of running a seti like
> >
> > project to reduce this data. Eventually we will have a well
> > understood
> > pipeline. It will be a big project to run all the data through it.
> > I am
> > prepared to do this myself and just set up a row of computers and
> > shove CDs
> > through them. But perhaps it would be fun to spread this around to a
> >
> > team. I know I brought this up before and got a good response. Now
> > I
> > could actually ship out data for analysis.
> >
> > The start would be the engineering run data. This will be 500-600
> > CDs, of
> > order 30,000 images taken at 7.5 degrees N covering all hour angles.
> > Even
> > if the calibrations are not the best possible, there should be lots
> > of new
> > variables in this data. My bet is that there are a few thousand that
> > have
> > not been measured before. The engineering run contains both long and
> > short
> > term data so we should find a wide range of periods.
> >
> > Let's discuss how we might do this.
> >
> > Tom Droege
> >
> >
>
>
> =====
> Chris Albertson
> Home: 310-376-1029 chrisalbertson90278@yahoo.com
> Cell: 310-990-7550
> Office: 310-336-5189 Christopher.J.Albertson@aero.org
>
> __________________________________________________
> Do You Yahoo!?
> Great stuff seeking new owners in Yahoo! Auctions!
> http://auctions.yahoo.com
>