On Wed, Dec 01, 1999 at 07:31:52PM -0500, Greg Maxwell wrote:
>
> Because of the nature of the gunnplexer, it is half duplex and collision
> detection is NOT easy. It also can only be implimented between two
> points. It would probably make a lot of sense to impliment a simple
> token passing (i.e. trade a token with every packet, keep it for x ms if
> you have nothing to send, only send when you have it, if you dont get it
> back for 2x ms consider the token lost and regenerate it) as a hack in
> the ethernet driver.
Hmmm, then why not just implement token ring? For one thing, for a
multi-drop wireless LAN, it would seem to be an interesting solution.
In fact it seems to me (thinking about it for the past 30 seconds) that
the radio LAN would serve as a virtual MAU (Media Access Unit). Only
trouble is that token ring testers and familiarity is much lower than
Ethernet, but I think Token ring would lend itself well to such an
application. The transmit and receive pairs are always split as the
network is always an electrical ring from one station to the next and
is a logical ring as well. There would of course be many details to
be resolved, but it may be worth looking into. I'm not a token ring
engineer but have gained a working familiarity with it in the past 8
years or so as a technician.
Another interesting possibility is that most token ring interface boards
will work at either 4 Mb/s or 16 Mb/s so experiments could start slow
if that were more economical.
Just food for thought from a guy on cold medication...
73, de Nate >>
--
Packet | N0NB @ WF0A.#SCKS.KS.USA.NOAM | "None can love freedom
Internet | [EMAIL PROTECTED] | heartily, but good
Location | Wichita, Kansas USA EM17hs | men; the rest love not
Wichita area exams; ham radio; Linux info @ | freedom, but license."
http://homepage.netspaceonline.com/~ka0rny/ | -- John Milton
