[PLUG] Personal Telco Field Day 2006 report

[Russell Senior]

The Personal Telco Project held a Field Day[0] yesterday in which it
practiced its ability to construct long-distance wifi links.  By
Slashdot standards these were pretty short ones, but for most of us
doing the practicing they were pretty long ones.

The Field Day has been an annual event of Seattle Wireless[1].  This
year we were invited to participate with a coordinated event here.  In
the end we discovered that Seattle Wireless had decided to put off
their event until October.

On Wednesday evening, a group of PTP activists hiked or biked to the
top of Mt Tabor to scout out sites.  I was surprised when, unaided by
special antennas, I was able to intermittently connect about 2 miles
to one of our nearby nodes, the Hawthorne Hostel [2], with my laptop.
With a viable link to the outside world, we considered other
destinations.  One of our group, Tamarack Birch Wheeles, works at
Portland State University and has access to the roof of the Ondine
building, which we could see clearly from our perch on Mt Tabor.
Another feature visible was Council Crest, which has good access by
vehicle, so we hopped into Tamarack's car and headed over to scout
there as well.

Council Crest has good views to the northeast, less good to the east,
but if you stood in just the right spot, in a gap between a couple of
trees, through binoculars you could make out the street light above
the spot where we'd been standing on Mt Tabor.

Over the next couple days, scurrying ensued, as we had not really been
preparing or planning very much since we first found out about this
event about 6 months ago.  I worked on getting some of our two-radio
outdoor enclosures[3] ready, assembling needed high-gain antennas and
other gear, considering power supplies, etc.  Then, at the last
minute, we discovered two things: a) that Seattle Wireless had
postponed their end, which took some of the pressure off; and b) that
due to family obligations, I (aggregator of most of the gear) wouldn't
be able to get started until late morning on the day of the event.  Oh
well.  We carried on as best we were able.

When I arrived on Saturday morning at about 11 am, Tom Higgins,
Tamarack and Don Park had set up near the reservoir on Mt Tabor and
were scouting out signals from a vehicle-mounted yagi.  This location
was further down the hill from our original site.  Tom could hear the
Hostel node, but not associate, which might have been an issue either
with the location, or perhaps the radio card or associated drivers.
Meanwhile, I set about configuring our Metrixes to make sure we had
them working correctly at close range before attempting the longer
shot to Council Crest.  Tom Higgins had brought a nice marine battery,
from which he operated an inverter which allowed me to keep my laptop
charged up, as well as supplied the Power-over-Ethernet injectors to
energize the Metrixes.

After about 90 minutes of fiddling, during which Sam Churchill
arrived, we had the Metrixes ready to go and configured for the most
interesting Mt Tabor-Council Crest link.  Tom and Don took their
Metrix, Sam's tripod-mounted 24 dBi dish antenna, and an omni with
them to Council Crest, while Tamarack departed for the Ondine and Sam
and I headed for the top of Mt Tabor.

The upper loop at Mt Tabor is beyond a locked gate, requiring us to
cart our gear up on foot.  Having anticipated this, I had brought
along a jogging stroller.  Sam and I parked at the gate, loaded up the
stroller with about 50-75 lbs of gear secured with gaffers tape, and
pushed it through the gate and up the paved road about a 5-10 minute
walk to the top.  Arriving on site, we set about assembling a
non-penetrating "tripod"[4] we had borrowed from the Mississippi
Grant Project stores.

Don had brought along FRS radios for the event, and we were able to
maintain voice contact between Mt Tabor and Council Crest while we
were getting set up.

At the Mt Tabor end we attached two panel antennas to the mast.  One
was a 19 dBi loaned by Sam Churchill, pointing at Council Crest.  The
other was a 13 dBi, pointing at the Hostel.  We discovered a slight
problem with the mast mounting hardware, which consists of machined
aluminum brackets, on one of the antennas.  The bolts that hold the
opposing brackets together around the mast were inexplicably not long
enough to accommodate the diameter of the mast.  We swapped bolts with
the other antenna so that the more critical 19 dBi antenna had the
better mounting and then jury-rigged the other less-critical antenna
with one bolt and some gaffers tape.

All this activity, with the odd, un-park-like gear, attracted a fair
number of questions from passersby.  One gentleman approached and
started to ask some questions, and it soon became apparent that he was
there "on purpose".  Jay was in town from his home in Seaside for
another event.  He'd heard about our Field Day from a friend of his
who lives in Arizona, who had suggested that our thing might be worth
checking out.  While Jay got filled in on some of the details, he pitched
in and helped as we assembled the final bits.

On Mt Tabor we didn't have a nice marine battery and inverter to use
with the PoE injector, so after mounting the Metrix to the mast, we
opened the enclosure and applied 12V to the internal power jack and
device sprung to life.

We used channel 1 on the longer link, since the Hostel's node operated
on channel 11.  I had configured the Mt Tabor node radio to AP mode
with an SSID of: www.personaltelco.net/tabor.  I logged into the local
Metrix from my laptop via ethernet.

Right away, we had a solid signal to the Hostel, and I was seeing what
our driver (madwifi-ng) reported as "associations" from the Council
Crest Metrix and one other, which turned out to be Tamarack's Apple
laptop from the Ondine.  However at first, neither of them were
_actually_ associated.  As we started fiddling with the gear to get
the Council Crest link working, Tamarack got his Cisco 1242ag device
out.  We fiddled a little with our antenna aim, which didn't help
much.  Then, ultimately, the Council Crest crew decided to try
flipping polarity of the antenna.  We'd set up vertical polarity at
Mt Tabor, and Council Crest initially thought they had as well (I
think they were just confused).  They flipped the tripod over 90
degrees, and BAM, association!  

We spent another 15 minutes or so getting some minor NAT issues and a
little local DNS problem worked out, until I was able to authenticate
with the NoCatAuth captive portal software at the Hostel node, and
then Council Crest was online!  Tom got on the IRC channel, made a
Skype call over the link, and was generally just pleased.

Afterwards, we repointed the Mt Tabor antenna about 5 degrees to the
right to hit Ondine, but despite polarity flipping, could not
establish an association there.  We decided, as the Ondine roof was no
doubt quite warm by this time, it was not worth trying to debug on the
spot.  We called it quits and began tearing down the gear.

The link to Council Crest was about 5.6 miles (9 km), according to our
GPS coordinates and an online calculator.  The radios at either end
were ~19 dBm, with 24 dBi and 19 dBi antennas at the end points.  We
had some modest Fresnel zone interference at the Council Crest end.
The link to the Hostel was about 2.0 miles (3.2 km), which we
connected to with a 19 dBm radio and 13 dBi panel.  The Hostel's gear
consists of an unspecified Cisco outdoor AP and antenna originally
donated by Intel and installed about 4 years ago.

We were pleased with our success, modest though it was, and resolved
to try something slightly more ambitious next time: perhaps an
additional link to Rocky Butte, and/or (if we can do it before winter
weather sets in) a longer link to the top of Larch Mountain, about 25
miles to the east.


NOTES:

[0] http://wiki.personaltelco.net/index.cgi/FieldDay2006
    http://www.personaltelco.net/gallery/fieldday2006

[1] ttp://www.seattlewireless.net/FrontPage

[2] http://wiki.personaltelco.net/index.cgi/NodeHawthorne

[3] These were Metrix Mark 2 devices, borrowed from the Mississippi
    Grant Project for the day.  They are waiting to go on
    yet-unidentified roofs in the Boise Neighborhood of North
    Portland.  One of the devices we used was recently reconstructed
    from surviving parts of our vandalism incident early this month.

[4] http://www.personaltelco.net/gallery/missipcommons/DSC01030
 
    This is a photo of one of the "tripods", or "sleds".  Since this
    was a temporary installation, there wasn't a lot of wind, and it
    would be attended continuously, we bypassed the usual cinderblocks.

-- 
Russell Senior, Secretary
russell at personaltelco.net