SS and AR

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

Dewayne, wa8dzp responds to n6rce:
>
>      I agree with Glenn's position in general.  However, I feel that there are

> certain "mission" specific requirements where omni-to-omni is the best choice.


Absolutely. This was one of the points I hoped me made in the two CNC
papers. As a start, grouping operation into three catagories seems to
be reasonable:

1) omni-omni      (perhaps <= 50 kbps and primarily below 1 GHz or so)
2) point-omni     (maybe 50 kbps-couple Mbps or so, low microwave freqs)
3) point-point    ( >1 Mbps, short enough wavelengths that the
                   transmitted energy can be focussed with antennas of
                   acceptable size)

Amateur resources and physics fundamentally determine the kind of performance
possible with each. High data volume, high speed and wide area demand
migration toward (3) but there will continue to be a vital need for (2)
and (1).
  Hubmaster is a member of (2,) the intermediate performance category.

  Increasing performance requires increasing coordination and cooperation.
Some important amateur applications, however, require that operation
with low or at most medium coordination be possible. For example,
emergency operations may not permit both ends to be point-point.
High information volume and speed may require it anyway though.
(on-the-spot TV news has a van with a directional antenna pointing at a
hilltop receiver which relays the video to blood&guts-news-@-11 at the
station.  For DX performance they may have to send someone to the
hilltop to point a directional at the van also(full pt-pt), for short
range an omni or gain-omni (pt-omni) may suffice).

I see coordination as the fundamental challenge for development of an AR
network.

(de Kevin, n6rce)
> >    However, since SS isn't channelized, your either going to have to
> >    develop a protocol to handle 500+ hams all on the same chipping
> >    sequence, or issue spreading sequences per group.  Again,
> >    coordination and cooperation is going to be the real problems,
> >    rather than get the technology to work.
>
>      I can't agree with you more.  We are really going to need better
> coordination and cooperation in amateur packet radio then there has been (at
> least in Northern CA) to get this type of technology into service in the
> amateur community.  Proxim has addressed the channel problem somewhat by
> implementing switchable channels as part of their design.  It won't however be

> able to handle the case where there are 500+ hams using these units.

Also it should be recognized that spread spectrum provides no free lunch.
While it doesn't require prior coordination in the same way that
frequency domain multiplexing does (channelization) and therefore doesn't
have a hard upper bound to the number of users (the number of channels
you chop the spectrum up into) it does deteriorate as users are added.
  Each additional user raises the noise floor for all others. If there
is, say, 20 dB of spreading then two equidistant stations using different
channel switching will produce -20 dBC unwanted power(noise) in the others
de-spread receive i-f. If the receiver requires, say, 10 dB of despread
S/N to demodulate (pretty optimistic) to an acceptable BER then things
collapse when the "other channel" user is 10 dB (1/3 the distance) closer.
As additional users share the channel each raises the collective noise
floor.
  Rather than reaching a clear limit to the number of users, as does FDM
with a set maximum number of users per sub-channel, SS sharing continuously
degrades channel performance.  It is probably a good deal like the way
omni/omni fails (witness the 145.01 MHz syndrome).  I suspect that
congestive collapse is a pretty likely possibility for this scenario
unless careful preventative measures are taken too.  This is also
somewhat like the way Ethernet appears.  For very lightly loaded systems
it looks like multiple users each get 10Mbps on a 10 Mbps channel.  As
loading increases the truth becomes apparant.  Reminds me a little of
inflation and the national debt too (:>)

  BTW, the 904 radios show BER in the 1e-6 territory for around -90 dBm
input as reported in QEX last spring.  The -75 dBm was an interim number
for the #2 ("1st pilot") radio with 500 MHz mixers in the 900 MHz spots
and untweaked filters.  Consequently overall gain was too low.  The
first radio, implemented for a 2 MHz channel and capable of supporting
500 kbps worked about as I would expect.  I'm in process of reselecting
filter component values for 256 kbps maximum speed and 1 MHz channel
size.
  Any of these simple FSK radios using the MC13055 or MC3356 chips seems
to want C/N in the low 20 dB area for 1e-6 BER.  This compares pretty
reasonably with the wa4dsy modem(MSK) which appears to get by with 2-4 dB
less.

73
Glenn n6gn