As has been pointed out, topo maps and visual LOS are only so
good.  If you want to know what your link will be like, the only
way is to test it.  Unfortunately.

Back in 1995-1996 when I was with PG&E's R&D lab, we built out a
wireless sensor network in Napa to monitor for electric faults on
the overhead lines.  We installed the master station with a high
gain omni, and then did our full site survey to position the
sensors.  The master station was on a hilltop chosen for good LOS
to the majority of the area monitored (and the fact that it had
infrastructure to support a readio installation).  That summer
and fall were the most fun I've had working... 4x4 truck driving
around Napa meeting PG&E troublemen with bucket trucks and taking
my notebook up for a radio measurement to the master station.
What we discovered was suprising.  Many of the locations that
topo maps suggested as good were in fact not all that good.  The
real world intrudes.  :)

We were particularly suprised at the alignment of the high speed
link back to the PG&E WAN.  Of course, in those days, high speed
was a 64kbps syncronous link!  Anyway, the actual wire mesh
semi-parabolic dish had to be pointed a little off direct facing
where the "mirror flash" indicated to get the best measured
signal strenth.  Again, the real world...  

There is no substitute for actually doing the real testing.  
Certainly, do the planning to reduce the chances of total
failure, but be prepared to modify your thinking and plans based
on the reality of the installation!

In your case, given the expense and pain of the tower
installations, perhaps renting bucket trucks at both ends might
be advisable.  You may want to go look at the trees too, and make
sure they have leaves when you do the tests.  Otherwise, spring
may come and give you a nasty suprise.

Greg

PS yes, that spread-spectrum radio network was linux powered :)









--
general wireless list, a bawug thing <http://www.bawug.org/>
[un]subscribe: http://lists.bawug.org/mailman/listinfo/wireless

Reply via email to