DIARY #20
THE MULTIPLE WIRELESS-DATALOGGER
RIDDLE
July 26th, 2000
One major problem has yet to be solved satisfactorily.
To explain it requires some technical background.
The historical way that Campbell Data Logging
- based data collection has worked, has been:
1. A copy of PCW208 Software runs in a Windows 95/98
computer. It can address one serial port, such as COM1, at
a time. It cannot access Ethernet or TCP/IP enabled ports.
2. The most universally used Campbell Data Loggers
- CR10, 21, 23X's have a single Serial Port for I/O on them.
3. The removable Memory Modules that work with the
data loggers likewise have one serial port.
4. To recover data from the data logger and/or the
Memory Module the user physically brings a laptop computer
running PCW208 software to the proximity of the Data Logger
and/or Memory Module, or brings one or both of them to the
vicinity of a workstation running PCW208 software.
5. One data logger, or memory module is attached
by serial cable (and either with Serial to DCE, or Optical
Isolation Interface) to the computer through a Serial Port
- usually COM1 - and the software is commanded to 'Connect.'
6. When the software, through the Serial cable detects
a Data Logger after sending a connect string and getting a
response, it then links up to the logger, and all actions
thereafter take place between the PCW208 software and one
- only - data logger.
7. The software does not know which data logger,
by identification, it is connected to. It deals with the one
in front of it. If the user connects two data loggers or memory
modules to the same computer it is up to him or her to save
the downloaded files in some identified way. The software
does not do this.
8. Thus Campbell Data Loggers, and PCW208 software
are designed to work with one, physically attached, system
at a time, only. There is no 'intelligence' in either the
loggers or the software to deal with two data loggers simultaneously.
The Freewave DRG115 series radios can be
configured to talk to many other radios - in a master, slave,
or repeater, relationship at once. The DRG115s can be configured:
1. In a point to point configuration.
2. In a point to multipoint configuration.
3. And the newest radios (Version 6.1 on) can be
instructed to use a 'Call Book' or in the newer 'Network ID'
Mode.
4. When in the Call book mode, each radio has to
be told what other radios it is connected to, by serial number.
So the 'radio' has a number which can be read from the outside.
5. In the 'Network ID' mode, no serial numbers are
used; so if one radio goes down it is not necessary to reprogram
all Call books on all radios.
The problem comes when two or more Data Loggers
are connected through one Master radio via the serial port,
to the PCW208 software, and the user command 'Connect' there
is no way for the software, or the radios, or the attached data
loggers to know which data logger is desired by the user. It
will simply arbitrarily connect to the first (fastest) responding
data logger.
The only way Campbell has recognized, as
needing a Data Loggers ID, is in (a) Cell Phone operation or
(b) 'Call Back' mode, where the Data Logger initiates a 'call'
when a data event occurs.
In the Cell Phone mode, the telephone number
of the Cell Phone attached to a cell phone modem, which is attached
to the Data Logger, is the only ID for that data logger.
Installers of Freewave Radios in other situations
discovered that it is possible to put the Freewave radio ID
in the software configuration PROVIDED that the Freewave is
in Call Book mode, and by that 'radio ID' method, permit the
user to call up a specific data logger.
But when the radios are put in the much more
effective 'Network ID' mode, telling the software to talk to
radio 571-9999 does not work, for when 'connect' is called,
the radio does not recognize any radio ids.
One Works,
Two Do Not
Thus, when the El Verde Station PCW208 software,
connected via the Master radio to the El Verde Tower CR10X data
logger, it responded just fine. For there was just one data
logger out there online to call.
But when a second data logger, such as at
the Bisley Tower, gets its data logger wirelessly connected
also, and both are reachable through the one Master radio, when
'Connect' is commanded, the radios will broadcast through ALL
radios, the connect string, and either BOTH the El Verde, or
Bisley, or neither (conflict), will try to respond, or some
unpredictable results will occur.
If we put all the radios into Call Book mode,
then we could put different calls to different RADIOS by id,
in the software, and it would work. But when some of the data
loggers are in very hard to get to places, such as on high towers,
or worse, inside weather-sealed boxes high on posts such as
on the mast at El Yunque, it will take a major, coordinated
effort, to reconfigure all radios in the net AND the PCW208
software if one radio goes down. So we do not want the Call
Book mode to be used, but the Network ID (no identity of radio)
mode. The Call Book mode does not scale for remote field work.
After many consultations with both Campbell
and Freewave, we determined that one or the other of them has
to made a software change. Preferably Campbell. The change would
be rather trivial. In the Edlog programing language which is
compiled to run in each data logger, the string and function
'Logger ID =' should exist. Thus an arbitrary logger number
could be entered, and thus stored in the data logger. Then,
a corresponding menu choice should be programmed into PCW208
software in their next version. 'Logger ID =' Both would then
be programmed, the data logger permanently, the PCW208 software
whenever used.
Then, when the user decides which data logger
is to be connected to, he or she enters the number into the
PCW208 software, and when 'Connect' is commanded - whether it
reaches all loggers via multi-point radio, or later through
Internet IP connections - the matching one only will respond.
Kludgy Fix
for Puerto Rico
Meanwhile, based on the only solution that
does not require Call Book radio addressing as the means to
differentiate between data loggers over multi-point wireless
links, we will program the data loggers to use the same techniques
they now use to control Cell Phone access.
We will allocate 20 minutes of every hour
to one data logger. Thus all three primary Weather station Tower
loggers in El Yunque forest will be available within one clock
hour. The data logger will be programmed to 'turn on' - literally
- the radio it is attached to. (Which technique was used to
save battery power with cell phones that are used for data collection
in areas served by cell phones.) Since the other two radios
will be 'off' when a PCW208 user commands 'Connect' during that
20 minutes, that one logger will respond, since it is the only
one 'on.' Then the first logger will turn off its radio, and
the second logger will turn its on. And so forth.
This technique has been used in a research
project that has 15 data loggers connected. And it is programmed
to do 'automatic' downloads of data, rather than manually accessed.
This will work, but it undercuts the one
advantage of 'always on' radios and data loggers - getting data
from any logger when, and from where (over the Internet) it
is desired.
So the short term solution is a kludge. The
long term solution awaits changes made by the data logger companies,
as they realize that wireless access changes the assumptions under
which their equipment has operated in the past.
NEXT
PREVIOUS