david@ms.uky.edu (David Herron -- Resident E-mail Hack) (02/18/88)
We have a local project which is in the middle stages of development
and bears some similarity to the Stargate project, so I was curious
at some of the algorithms used in stargate.
The project is being run by the state of Kentucky and is intended to
connect grade/middle/high school teachers (probably only the math
teachers, but I'm not sure) across the state into a bulletin board
type system so that they can become better teachers. I'm not involved
in the day-to-day thinking on the project, so I don't know the details.
But I have been involved from it's beginning a year or two ago, and it's
likely I'll be more heavily involved as it moves into its' software
development stages.
The medium we will be using is a line in the VBI interval of the
states' educational TV network. We are apparently somewhat unique
in that one of the requirements for Kentucky Educational TV (KET)
is that their signal reach all corners of the state, so the first
problem is solved -- namely reaching the whole state.
We are planning on using UUCP type stuff (using the UUPC software)
to get news/mail back to a central site.
We have developed some circuitry which is fairly simple to build,
yet allows 120 chars per second (2 per VBI line) to come out of
the TV signal. The circuit is simple enough that someone could
go down to Radio Shack and build it from parts found there.
The part which hasn't been finished is the protocols for transmitting
news over the tv signal. That is -- it's obvious that there needs
to be some redundancy in the information we transmit, but when
do we repeat and how often? How do remote sites tell the central
site that they missed an article? Note that they may not even have
gotten ANY of the article and therefore don't have any notion of
what articles they are missing. One of the protocols I envision
is something along the lines of the newer uuencodes -- that is,
have a length indicator at the beginning of the line and a checksum
at the end. With such a protocol I could see a higher level of
it realizing that it had an incomplete article and if it had enough
of the article it would have the article-id and be able to request
it to be re-transmitted from the central site.
(Oh, I'm using Usenet terminologies because we'll probably be using
Usenet software and may even be feeding a full Usenet out to them).
I have a vague memory reading that Stargate repeats each article
4 times a day... Is that merely because 2 megs per day at 1200 baud
can be repeated 4 times in a day? (That's one of the calculations
I did recently -- the same calculation showed that we'd likely be
unable to effectively send Usenet out over KET because KET only
broadcasts 11 hrs a day or so).
We're going to be entering the testing-of-the-hardware stage in
the next month or two or three ...
How small a computer does it make sense to support? The decoder
box merely produces an ascii data stream out an rs232 port. Just
about every computer has an rs232 port (or can have one). Do we
just sneer down our noses at people that have C-64's and say
something to the equivalent of "Get a *real* computer"? But obviously
something like a C-64 couldn't *dream* of being able to hold a
full Usenet feed. Possibly some sort of "broadcatch" program
(read _The_Media_Lab_ for a definition of broadcatch) could work
for them, but they still have an ~360K limit. Also, their computer
would be pretty much tied up all day long receiving news -- but
I don't see many ways around that. (I know -- tell 'em all to
get Amiga's and make news reception a background process :-)).
Any ideas?
--
<---- David Herron -- The E-Mail guy <david@ms.uky.edu>
<---- or: {rutgers,uunet,cbosgd}!ukma!david, david@UKMA.BITNET
<----
<---- It takes more than a good memory to have good memories.