MAC sublayer protocol

[Glenn Elmore <glenne@hpnmdlc1.hp.com>]

Barry writes:
> likely the RF problems resulting from shared usage.  However, the
> application of full duplex I was addressing is CSMA MANs with restricted
> coverage.  They don't need to be (and shouldn't be!) on mountain tops or
> TV towers, so the possibilities of finding a reasonably clean RF site are
> greater.

Having already mentioned some of the difficulties in restricting
coverage I'll say that I agree strongly with the principle.  In
selecting sites for hubmaster we are trying to find locations which are
*just* LOS to all the intended users and no higher.  It turns out that
with the approx.  10 mile max cluster radius and users with the 16 dB
antennas that considerably less than 10 watts is ample. Recent
measurements show over 40 dB of excess C/N with 10 watts. This can reduce
cost at the same time it allows better spectrum reuse by other
not-so-distant clusters.  A useful and interesting possibility is to
have the server/hubmaster/repeater not located at the physical center of
things.  I think it turns out to work pretty well if that server is
located off away from the center of cluster population...perhaps part
way up the side of a mountain bordering a valley located user base.  If
the server uses a directional antenna with beamwidth chosen to just
include the subtended angle of all users the antenna gain makes up for
the extra link DX.
   In any case the thing wants to be as low as possible to do its best
at "getting the power/information directly where its supposed to go and
no where else".  It has struck me that this is also a pretty good
philosophy for routing table entries.

> Although the spectrum coordination issue may be more confusing, the use of
> crossband full duplex allows some room for creativity in dodging the
> problem that Fred mentioned, where all of the coordinated in-band duplex
> pairings tend to be locked up by the voice repeater crowd.
>
  <deletion>
>
> same bands as the first one.  The new repeater can be multicoupled to the
> existing receive and transmit antennas, and no duplexers are required.  In
> fact, in the case of the DSY modem, it appears quite feasible to do the
> multicoupling at the 28 MHz IF.

Before I sold my dsy modems (to help finance higher speed radio
hardware) I had considered doing something like this.  As an old
posting, "whither layer 1?", indicated, the price/performance of 1200 MHz
is pretty good.  Since there were two high speed digital allocations 50
MHz apart I had thought of running the dsy transmitter at 78 MHz or so.
Putting it above the receive i-f and keeping signal levels low prior to
upconverting gives a pretty good chance of FDX coexistance.  I even
think I heard someone mention success at running the Tx continuously
on-channel at 28 to avoid keyup delay with success so this trick would
likely work fine.
  By chosing transverters with high or low side LOs you can pick whether
you transmit high/receive low or vice versa.  With that much frequency
separation I'm pretty sure that some cheap interdigital bandpass filters
(more hobby tubing and PC board) could allow FDX operation.  The LO I
showed in the HR Mag.  microwave design series is simple and can easily
operate in this range.  I'm currently using it in my 1296 transverter in
the shack.

  About the same time N8XXX (call?)  had suggested the "hexnet"
configuration.  I noticed that by using three different FDX pairs it
would serve the multi-color map (freq reuse) issue at the same time it
used current hardware.  As an improvement to performance, PIN switching
the TX to one of 3 directional antennas improved system margins/allowed
lower power.  The receiver ran gain-omni serving each of the three dsy
channels with the one receiving converter.  If anyone is interested I
can e-mail the arrangement.

> would be fairly easy for users to switch channels, by simply switching
> crystals in their DSY modems (a simple synthesizer design might even be
> feasible).

With hexnet I think 6 crystals covered all possibilities...

>             Intriguing concept, no?

I thought so.

73
Glenn Elmore