DIARY #14
DATA FLOW
CHOICES; DATA LOGGER LIMITATIONS
On July 7th we held an important meeting
between Tim Kratz, PI, Paul Hanson, Data Manager, Tim Meinke,
Technical Assistant at Trout Lake, and myself as Wireless Project
PI. Margot attended also.
For the first time the practical problem
of changing over the data collection and flow system, from either
the manual 'retrieve the storage module' or the cell phone dial
up method, to a wireless collection, and Internet transmission,
system was addressed by the LTER Staff. We of the Wireless Project
cannot install radios on the Trout Lake data loggers until that
end-to-end new data flow system is established. Our mandate
stops when we hand the PI either an RS232 connector, or an Ethernet
RJ11-tipped wire, and say "Data is now flowing out the end of
this wire; what do you want to do with it?"
The Multiple Data Logger Addressing
Problem
The basic problem, which affects almost all
research and data collection sites which use Campbell Scientific
Data Loggers - CR10X, CR21X, CR23X - which are very, very, widely
used comes from several basic facts:
(1) The assumption underlying a Campbell
Data Logger is that it will be deployed to the field, connected
to a series of sensors whose data is stored within the data
logger's memory, or external, detachable, memory module, in
accordance with unique programs written and compiled in the
logger. Then it is expected that the researcher goes to the
field and retrieves the logger, or the memory module (replacing
the module as the logger continues to log). Then, using a short
length of RS232 Serial Cable - and an Optical Isolation Interface
device - the researcher connects up the Memory Module to the
PC, and runs the PC208W Campbell Software in a Windows machine.
The software in the PC connects up with the software in the
Memory Module, and the data is transferred from one to the other.
As often as not it is saved to the PC's hard disk. And perhaps
printed. Or, sent as individual text email to a distant researcher.
(2) The design of the Campbell assumes one
computer, one data logger, a serial cable in between, and the
PC208 software interacting with the software - and stored data
- in the data logger. No connection to the Internet, or even
an Ethernet port to the Internet through a LAN. Only one data
logger attached at a time to the computer. There is NO identification
of one data logger from another. So if two data loggers were
connected simultaneously through the same Serial Port of a PC,
the standard Campbell PC208W software cannot, as currently written,
distinguish between the two! Yet that is exactly what we are
doing when we connect up more than one data logger to a single
computer at the research center via wireless!
(3)Something has to give. Either the proprietary
PC Campbell software has to be modified, along with the software
in each data logger, in order to give it a unique, numeric,
'identity' or some other switching method has to be used to
let the PC Software 'address' only one logger at a time, or,
physically, the radios have to 'switch off' access to the data
logger radios not being accessed. Unlike the Internet with its
TCP/IP protocols, Data Loggers are not yet IP addressable machines.
Only some radios attached to them - like the Aironets - can
handle unique IP addresses.
In the case of Trout Lake Station, only one
CR21X field data logger - the one on Sparkling Lake - will be
connected wirelessly at the beginning. Others will be added
later. And the Big Buoy system will have a singular connection.
So it is important for us to figure out how to solve this problem.
At the strategy meeting we recapped the way
the data from Sparkling Lake is handled now.
(1) The Cell phone system attached to the
Sparkling Lake CR21X is set up so it is called, long distance,
from Madison, Wisconsin once a day, during the brief periods
daily when the data logger software 'turns on' the cell phone
so it may accept calls. Data then is fetched from the logger,
by the PC208W Campbell software and is added to the Data Base
for the LTER, in Madison. That is the routine that is followed
daily.
(2) The Cell phone can also be called from
Trout Lake Station, but this is only done when there is a problem,
as reported by Madison. The two calls are independent.
(3) The once-a-day calls are driven by the
need to conserve the battery power on the raft from running
the cell phone constantly. And, to a lesser extent, the costs
of the long distance cell phone calls would mount up, if the
logger were accessed much more frequently.
I questioned whether or not, once it becomes
possible to fetch data continually, whether the data flow, storage,
and distribution should change. Tim Kratz felt that there is
little need for 'instantaneous' data from the logger - minute
by minute accumulation of data.
But whether or not, or how often the data
is fetched wirelessly, the reality is that it will be arriving
at the Trout Lake Station building, not in Madison, and must
be gotten to Madison if they continue to store it.
So Paul Hanson came up with a clever solution
that will work for the Sparkling Lake situation described above,
but may not scale to two or more data loggers connected to the
same serial port, and will only work if there is a Microsoft
Windows or NT machine at both ends.
The PC Anywhere Solution
The solution is to mount PC Anywhere on a
computer in Madison, which can access, via the Internet, a Microsoft
machine at Trout Lake which also has PC Anywhere mounted on
it, that is in turn connected out the serial port to the data
logger on Trout Lake, wirelessly.
PC Anywhere is a venerable piece of software
which permits a person on one computer to access another computer,
both Microsoft OS machines, and both running PC Anywhere and
'take it over' just as if one is at the console of the remote
machine.
In this instance, Paul would mount PC Anywhere
in Madison on an NT machine, access the NT Server at Trout Lake,
through the Internet, and, by driving the 'local' session as
visually as if one were at Trout Lake before the console, he
can 'run' PC208W software, which in turn links to the remote
data logger, where the wireless connection acts just like an
extended Serial Cable. He is prepared to set up that solution
soon.
That will work. But it requires that both
ends are running Microsoft powered machines - PC Anywhere does
not run on UNIX or Macintosh systems. And it does not deal with
the multiple data logger problem, linked over the same serial
port.
However, that will work long enough for everyone
to work out a much longer term solution, that scales for other
research projects, and multiple connected data loggers.
We decided to set up at Trout Lake Station
a test data logger, sitting next to the Internet linked NT Server
with PC Anywhere, and to exercise a wireless Serial Radio connection
between the Server and the Data Logger, which is being driven
from Madison by someone on an NT, running PC Anywhere. If this
proves reliable, without loss of data through all the links
after we have modified the Data Logger software to avoid the
cell phone routines, we can install the solution on the Sparkling
Lake raft and data logger, and the LTER can start solely depending
on the wireless connection.
But a better solution is needed.
