All
Radio Links Working
In
a short, two day visit to the Caribbean National Forest October
19th, 2000, we managed to get all three Campbell Data Loggers on
top of the El Verde, Bisley, and Pico Del Este Weather Station Towers
connected to computers in both the El Verde Research Center and
the Sabana Work Center, where Dr. Fred Scatena's data gathering
group from the National Forest Service work.
They
are not all working to their expected potential yet, but it was
with some pleasure at 5:30PM, Friday the 20th, that we got the final
'Connected' signal back from the PCW208 software in the Sabana Work
Center on the eastern edge of El Yunque forest. We will have to
do some work to optimize the performance of the 6 data radio network,
for as the data loggers that required two radio relays came online,
we observed reduced radio performance that affected data transfers.
And
we got one big surprise on the top of Pico El Yunque.
As
reported in Diary Report 18, we already had succeeded in getting
the Campbell CR10X Data Logger on the top of the 120 foot El Verde
Tower directly linked the short distance - perhaps 100 meters -
through the forest canopy to the computer room of the El Verde Research
Station. But at the previous field visit, we had put both radios
in the singular one Master, one Slave mode because only one data
logger was being accessed. That was easy.
But
now we had to apply the configuration techniques reported in #22
- which required the use of Radio Serial Numbers as pseudo 'cell
phone numbers' to get more than one data logger to respond individually.
And do it through two relay radios for the more distant towers,
one on El Verde Tower, and the other the Repeater Station Radio
we had installed on top of Pico El Yunque.
This
would require completely reconfiguring the three radios already
installed, and newly configuring radios for Bisley and Pico del
Este - as well as experimenting with one at the Sabana Work Center,
where, to our surprise, we had gotten a connection through very
difficult terrain to the radio on El Yunque from the work center
deep down in the forest. I knew that the configurations would be
very complex, and would have to work the first time when installed,
because the changes we would make would not permit us to know -
by simple, always-on, status lights on the Freewaves, that they
were communicating.
So following our own final advice in the configuration steps spelled
out in Report #22, we laid out all radios on the large central table
in the El Verde Center to get them all properly communicating with
each other in a 'bench test' - where we could observe the status
lights at all points - before carrying the configured radios to
their final locations, where both voice communications between the
towers and El Verde would be difficult or impossible, and trouble
shooting with laptops on top of the towers with the wet weather
conditions would be slow and frustrating. Just a little water in
any of the connectors could cause problems for days and weeks. I
had decided, in view of the raining weather on top of cloud covered
El Yunque, that it would be wiser to configure a radio in the lab,
and replace, rather than try to reconfigure the Repeater up on a
pole on top of the mountain.
Mike
Willett and I on the evening we arrived at Rio del Mar hotel on
the beach, did a quick test to see if the Repeater Radio on El Yunque
was still operating properly. It had been two months, and a Hurricane
had passed through. We wondered if the flat solar panel had been
blown away from its exposed position on the mountain top. Or that
other things had happened.
We
got a connect green light right from the balcony of our hotel room,
showing that the Freewave was still operating!
I already had reports from Andrew McFadden, that the data radios
that were connected to the El Verde Tower CR10X had worked flawlessly,
and through the hurricane, the past two months.
Configuring
After
buying two new solar panels from West Marine, in El Farjdo, and
some brackets, which we had ordered in advance, we arrived at El
Verde Station by 10AM. Fred Scatena, the chief Forest Service researcher
in the Luquillo LTER, could not attend, but, as I had requested,
he sent three of his Forest Service Assistants from Sabana to observe,
and start getting trained on the configuration and operation of
the FreeWave installations. And Andrew McFadden, technical assistant
from Dr. Jess Zimmerman's operation at El Verde and back at the
Ecological Institute in San Juan, was there. For in the long run
they would all have to be the ones who maintained the radio network
we installed, and to expand it, without our help. For this NSF Project
was designed, not only to 'model' and report on Wireless solutions
for data collection, but to pass the know-how torch to the technical
assistants at the research stations we worked with, so they could
do it themselves after we had departed. There are simply no local
commercial installers to call upon to do this kind of work. It is,
and probably will be, for a number of years, a 'do it yourself'
operation for research centers in remote areas.
So, for everybody's benefit, we ran the configuration as a sort
of Workshop, explaining as we went along, and demonstrating how
the radios worked with computers and a CR21X that Dr. Jill Thompson
loaned us for the bench testing.
It
took most of the rest of the day to get the radios configured right,
even following the steps shown in Report #22. It was, to excuse
the French, a bitch. Part of the complexity came from our attempts
to make it possible for Carlos Estrada, the Data Manager at the
Sabana Work Station, to access the two Data Loggers at Bisley and
Pico Del Este - the ones they are responsible for in the first place,
from that location, while Dr. Zimmerman's researchers only download
the data from El Verde - which it is responsible for. Otherwise
the El Verde Station would have to download the data onto disks
and get them to Sabana. We needed to match the division of labor
already established, or else the El Verde Station would be the only
point from which the data in all three weather stations could be
downloaded, even though two of the weather stations data collection
were the primary responsibility of Fred Scatena's Forest Service
Sabana research center.
We
found we had to do one basic technical thing - remove the 'Security'
flags in the radios, which insured only one specified 'Master' radio
could access the data. Alternately, the radios at El Verde and Sabana
would have to act as 'Master' radios, to get their respective data.
So, in theory, someone else, with a FreeWave radio properly configured,
and in range of the El Yunque repeater radio, AND a copy ($300)
of PC208W software to run on a connected PC, also properly connected
(with specific serial numbers of the radios locked up in the cases
in the forest) could download data from any of the stations. We
figured the odds of someone doing that, given the value of recent
weather data, to be extremely low. Not quite like breaking into
Microsoft's Corporate Network to steal source code!
We
finally got all radios working on the bench in the late afternoon.
In particular, the way the Freewaves have to be set up to work with
multiple Campbell Data Loggers, only one data logger can be accessed
at once from one Master radio station through the PC208W software.
And we were uncertain whether access via two relay radios - thus
involving four radios in one download, would work at all, as it
must. Even a cell phone call back to FreeWave did not conclusively
settle the matter - only trial and error would. For FreeWave's Techs
had never tried it themselves, although there was no particular
reason why they would not work. The call did cast light, however,
on what it was going to take to get two 'Master' radios to work
in the network, one at a time (from the two field research locations.)
Much of the uncertainty, as well as 'proof' of working, comes from
the fact that when a Freewave is set up as a Point to Point Master,
driven by the PC208W software using the weird set up commands of
the form ATDXC05551234, the only way you are certain that the links
all the way through repeater radios are working to the end data
logger, is that the CD lights all go Green in the path, while all
other radios stay blinking, unconnected Red. And even then, only
if you have a physical data logger connected to the serial port
of the last radio, do they all light up, showing the ultimate correct
connection! (If there is no data logger, or other serial-answering
device connected to the last, Point to Point Slave radio, they do
not go Green, but remain in another red and dim red lighted condition,
and you are NOT 100% certain the links work).
Because
all the radios were in one room, as a miniature-distance duplication
of the entire network of 6 radios (their cross room power being
sufficient to link radios even without larger antennas), we were
systematically able to hook the one CR21X data logger to the three
Slave radios in turn, and test the total connectivity to the data
logger's responding software. With the Windows 98 machine with PC208W
software connected by serial cable to the Master radio, AND with
the 'Low Level I/O' window being opened next to the PC208 main program
window, we were able to both observe the lighting up of a given
radio pathway on the radios, and see the returned data, from the
dialing string, through the first CONNECT (which comes from the
first repeater radio, not the second one, or the data logger), and
finally, as the PC208W software indicates its 'CONNECTED' lights,
one can observe a continuous flow of data from the logger all the
way back through the chain to the PC, even though none is being
captured or downloaded.)
I cannot stress enough, how important it was to do the whole thing
on a bench test, with all radios configured and connected, before
installing a single one, remotely. Had we not done this, we would
have failed in our field installation, and found ourselves troubleshooting
with laptop computers up on towers in the rain for ???? For even
when the bench tests worked right, issues of latency as the radios
switched, and less than 90-100 percent radio strengths, coupled
to the very touchy Campbell PC208W software which was never designed
to work with long, or unpredictable, delays in the interactive data
and control flows, could defeat a particular link.
Andrew
McFadden, demonstrating his increasing mastery of Freewave radios,
(by study of the manual, and having a spare pair of DRG115s to work
with) was a great help in my getting all the radios to work. Even
in one room it takes two technically savvy persons to get it working.
All
this, of course, is caused by the absence, in Campbell Data Loggers
and corresponding functions in their PC208W software, of any way
over a radio network to 'identify' a given data logger, except by
the identity of the radios connected to them. It is the radio tail
wagging the data logger-software link dog. Not, in my opinion, a
very satisfactory solution for the long run, or spread of the use
of serial radios in an extensive network of Campbell Data Loggers.
(I have made my views known to Campbell, that all they would have
to do in a future upgrade of their softwares would be to add a function
in the Edlog software to set 'ID=' to any arbitrary number, and
a corresponding setting in the PC208W software setup to match that
number for a given download session, and all this extraordinarily
complex, and extremely limiting of Freewave functionality - from
Point to Multi-point simultaneous use radios - to Point to Point,
fixed, one session at a time, would be avoided. And monitoring by
visual inspection of remote slave radios status lights while setting
up, or trouble shooting, would be much easier.)
By
late afternoon, Mike Willett, who had been working to get the Solar
Panels, connectors, SC932 interfaces, cables, in-line fuses, and
batteries charged and ready was ready to go up El Yunque with Andrew
to replace the Repeater Radio at the top of the main mountain, with
a configured one.
Andrew first reinstalled the reconfigured Slave-Repeater radio on
the El Verde Tower nearby and its CR10X Data Logger, and we tested
that connection to the Master radio in the Lab before they set off.
It worked as expected.
Water
on El Yunque
There was driving rain, cold wind, and lightning crackling around
when Mike and Andrew got to the top of the mountain through the
locked gates to access the singular Repeater Freewave radio that
had been there for two months since we first installed it. There
was nothing to do but brave climbing the pole in the fading light,
miserable as it was, before darkness, or losing much time by waiting
until the next day. So Mike went up the pole to the 'waterproof'
box he had installed two months before, and opened it.
There was 6 inches of water in the bottom of the box inside! The
'waterproof' seals had failed (perhaps a wrinkle in the thin rubber
seal around the cover). There was no drainage hole in the box, so
the water just accumulated. And the FreeWave DRG115W - the 'waterproof'
radio model, had been siting in the bottom of the box, status lights
down, connectors on top, in standing water for over a month. And
it was STILL working properly! I decided to send a testimonial to
Steve Wulchin, CEO of FreeWave in Boulder. His waterproof radios
were indeed waterproof!
Mike
replaced the radio with the newly configured one, drilled a drainage
hole in the bottom of the box, and positioned the radio so that
its status lights could be seen from the ground, outside.
The
Solar Panel was still in place, doing fine. The only thing they
did not get to check was the voltage level in the West Marine Gel
Cell batteries - to see if it was keeping up a charge, given that
the solar panel was in constant mist and clouds. We must do that
later.
So
a cold, wet team came down the mountain the end of the first day.
Three of the six radios, properly configured, were installed. And
data was flowing again between the computer in the El Verde computer
room, and the data logger on top of the tower, using the new 'dial-number'
configuration. There was no easy way, by status lights to insure
that the Repeater radio was working properly, when it was not connected
to any Slaves. We had to cross our fingers that the setup was right,
and that no physical connections were flaky, or wet. It's to Mike
Willett's credit, he did the job under nasty conditions and it worked
properly when he was done.
I was a bit discouraged by the news sent by Fred Scatena of the
US Forest Service, when we arrived in Puerto Rico, that other sections
of the Forest Service responsible for the permitted activities on
the top of El Yunque - a National Forest - refused his request -
even though the US Forest Service itself does environmental research
in that forest, the only Tropical Rain Forest under US Government
control -to grant the NSF funded LTER, a permanent permit for the
one self-contained and self-powered small radio, waterproof box,
antenna, and solar panel to remain in its present location for more
than a year. After which it would have to be moved to some 'other'
location on the top of El Yunque, perhaps on the same building the
Forest Service uses for its own AM radios. The trouble being that
that building in in the midst of the very heavy electromagnetic
environment of the Castle Communications dense antenna farm on top
of the mountain, several hundred meters away. We have no idea whether
the relay radio with its 1 watt of power connected to a 4-6 dB omni
antenna in the 902-928 MHz spectrum range will work in that 'other'
location.
We
will have to go through a whole new 'site survey' to find out. Which
we should do sooner, rather than later. And before our NSF Wireless
Test Project is done in Puerto Rico, or else it will fall to the
expertise of the Techs of the LTER to make the switch, rather than
we with considerably more RF expertise. Once again underscoring
the need for local university or government environmental project
technical staffs to learn how this stuff works from us, as we deploy
the radios. For moving, changing, adding, data radio links will
be an ever new job, locally. In this respect it is not much different
from the progressive need for local research project staffs to learn
a lot about their new and upgrading PC software, LANs, IP networks,
and all the computer, wired, and Internet technology being applied
to their science today.
Getting
data radios to work in the jungle is not Plug and Play!
Connecting
the Final Data Loggers - Pico del Este First
On Friday morning, the 20th, the Forest Service Team rendezvoused
at the El Portal gate at 8:30 AM- the route to the Pico del Este
- and, together with Andrew and Mike, set off to install a radio
at the mist-shrouded, but ground level weather station in the eastern
end of El Yunque. I stayed down the mountain, to run the connectivity
tests when they gave the word by - hopefully - cell phone. The weather
was better.
By
11:00 AM the work was essentially done on that site. While the cell
phone communications was spotty (we actually traded voice mails),
I made periodic tests from the computer inside El Verde Station,
where I had configured that one computer and PC208W software to
to fetch data from all three weather stations. The data logger was
a CR21X, older model.
I finally got a connect at 11:10. I left the system in the 'connected'
mode, observing continuous data flow on secondary computer screens
over the next half hour and varied the 'delays' in milliseconds
in the setups. The connection broke several times - generally after
a full two minutes of continuous connection. That might be bad,
if there were breaks during a download, given the way Campbell's
data loggers work. But it was serviceable for now.
What
this meant was that, with the data having to go through four radios,
two relay, we needed a more robust signal between the Pico del Este
site and the El Yunque relay radio. This would take fine tuning
later. Perhaps better aligned antennas, or ones of greater gain.
The
Final - Bisley - Tower Connected
While
the team came down from the mountain and prepared to go up the Bisley
Tower up the Bisley tributary, I made my side trip and preliminary
reconnaissance to Rio Espiritu Santos, to
begin planning for wirelessly capturing fresh water shrimp by video
later.
Getting
to the Bisley Tower, especially while carrying heavy batteries and
awkward solar panels, is very hard work. Made the more so by the
wet condition of the ground on a quite steep slope which turns slickly
muddy. It took a long time for the team to get to, and up the tower.
They
saw evidence of the work being done to study the rain forest, by
a Dutch team of researchers, who found bees nesting in some of their
plastic boxes on the tower they had failed to close.
By
2:30 PM Willett was ready for me to test a link to the Bisley Data
Logger - which had to go from El Verde Computer Room radio, to the
repeater on El Verde Tower, to the repeater on the top of El Yunque,
and then down to the Bisley tower radio, thence inside to the CR23X
data logger.
Once
again, I got a connection, but saw some dropouts. Less than at Pico
Del Este. Once again, there would have to be some later optimization
of the radio links.
So
with Bisley Data Logger connected, all three Data Loggers on the
three weather station towers in the El Yunque Rain Forest were connected
wirelessly. The first priority task set for me by Dr. Jess Zimmerman,
overall PI of the NSF Funded Luquillo Project was successfully accomplished,
if not finally completed.
With
the bad weather, we did not get any good pictures from the installations.
Connecting
from Sabana Work Center
Then I rendezvoused by 3:30 PM with the team coming down from Bisley
at the Sabana Work Center which is part of the Caribbean National
Forest support center. The data management building looks like this.
Putting an antenna on the flat roof would not be hard. A simple
tripod might do, unless we needed a tall mast, as we had installed
at Trout Lake in Wisconsin. Or something in between. The top of
El Yunque was definitely not in view. All we had to go on was that
we had got a connection from the vicinity of the Sabana building
two months before, when the Relay radio on Pico El Yunque was set
in multipoint repeater mode, and CD status light of any slave radio
would go green. Now that the Repeater radio had to be put in its
Point to Point Relay mode, we had no way to test the link, except
by creating a Master Point to Point radio at Sabana, reconfiguring
the PC208W software in Carlos Estrada's computer, connecting everything
up, and using what antenna we had (a 4 db omni) mounted on the roof,
to see if anything would work.
After
a good deal of work taking almost an hour and a half, paralleling
the previous day's session at El Verde, we finally got the desired
connection! We knew, because we had reached all three data loggers
from El Verde, through two relay radios, that the link from the
relay on top of El Yunque would work to Bisley and Pico del Este
radios and data loggers. We hoped it would be even a better connection,
because the Sabana Master radio would only be going through the
one Repeater radio on El Yunque.
But alas, this was not to be seen. The link worked to both remote
data loggers that were the responsibility of Sabana, but when Carlos
tested an actual download of 10 days data from Pico del Este - which
would usually take 2-3 minutes when he had the data memory module
right beside him fetched manually from the mountain - took over
10 minutes by data radio - indicating a poor connection. The radio
reported only 42% efficiency, where 65% is need at the minimum.
But,
since we did not have an 8 to 10dB directional Yagi at the site,
nor good quality LMR400 cable between the antenna and the radio,
we knew we needed to test downloads after we had a better connection.
So,
by 5:30PM, we had done all that we could. Carlos took command of
his computer, having learned how to use the radio. I turned over
a brand new DRG115R Freewave radio, antenna, power supply and miscellaneous
other things to him, for which he receipted. And we asked him to
perform tests to link to the data loggers, and keep a log for us.
Carlos
at his computer at Sabana.
That evening over supper, I instructed Andrew McFadden what he should
do before our next trip there, and authorized him to expend up to
$500 to buy whatever was needed to give Sabana the best possible
antenna link to the top of El Yunque, with a directional Yagi, tripod
antenna stand, post, and LMR400 antenna cable.
If that was unsatisfactory, we would have to try putting up a higher
mast for the antenna. Or, at worst, create yet another relay radio
location between Sabana and the top of El Yunque. We know it can
be done. And we have learned a lot about what it takes to set up
a radio network linking Campbell Scientific data loggers - the most
popular equipment for environmental scientists - in rain forests.
This
series of blow-by-blow Diaries should help researchers and their
project staffs realize what many of the hidden factors, and costs,
largely labor and time, are required to deploy wireless solutions
for field data collection tasks. And the reports should also help
manufacturers of both data loggers and radios improve their products
to permit much more extensive use of wireless for field data collection,
rather than rely, as in the past, on ultimately much more long term
labor costly manual retrieval of data from field loggers. For all
the NSF project costs over the past year to experimentally deploy
the initial radios in support of the El Yunque rain forest LTER,
the payoff will be in requiring less technical assistants required
to manually go into remote areas just to being back data modules,
and more use of technicians time to expand the number of data logged
sites by using wireless connectivity for recurring data collection
tasks.
We
left Andrew McFadden with a copy of the NL100 ethernet to serial,
and TCP/IP manual, in anticipation of the day, hopefully over the
next year, when full Internet IP connectivity is extended from San
Juan, and the Ecological Institute to the El Verde Research Station.
And we can begin to address the issues of very remote - from Puerto
Rico - scientists fetching data from the data loggers in the rain
forest, directly from their parent university locations - over the
Internet.
The final payoff use of wireless for field environmental, and biological,
research.
