MAC sublayer protocol

[uunet!aupair.cs.athabascau.ca!rwa]

In <9101151744.AA06967@hpnmdlc1.hp.com> Glenn Elmore <glenne@hpnmdlc1.hp.com> wr
ites:
>   fred k1io writes in response to Steve Johnson:
>>      [good stuff deleted for brevity]
>     If the repeater is located such that it *only* radiates to<->from the
>   same set of U users which would otherwise be trying to CSMA it alone I
>   agree.  BUT, very often the repeater is located at a high level site.
>   This means that it not only gobbles the spectrum from a much larger area
>   but also is likely to increase the number of users.  The result can be
>   that 10*U(or whatever) users now vie for the same channel and we are

Some trivial solutions:

1) put the repeater lower (saves $ too)
2) run lower power and/or more squelch (another $ saver)
3) okay, put up ten times as many repeaters.  Why assume
   a scarcity when none need exist ?

and a slightly less trivial solution

4) The above discussion assumes a "no-brains" full duplex repeater.
Ok, maybe it's doing some regeneration.  Now let's assume a few more
smarts:

    When the repeater decides that a packet might be arriving, it
    starts sending SYNC chars on the Tx side BUT it starts assembling
    the incoming packet at least until it has the calling station id -
    WITHOUT simultaneously retransmitting the incoming octets.  Then
    it looks up the calling station's id, and start repeating the
    packet if the call is on the approved list.  Otherwise, it fakes a
    collision on the output side.

The induced latency is only the time between the start of the SYNCs
and the end of the calling station id, which in a standard AX.25
packet is very near the head of the packet.  A little FIFO is all
that's logically needed.  If the packets are of a good size (a
given, since otherwise all the other latencies are going to kill us)
this won't hurt much, and if not - well, it's *still* better than the
alternative :-).

Mind you, I find the concept of a closed repeater to be one of the most
distasteful ideas in hamdom.  If the ham population is dense enough
(in the *spatial* sense of the word) to require this sort of
technique, I suppose keeping track of who can use the repeater would
be a bearable administrative overhead (sigh).  If the other sense of
the word applies, ghods help us.