SR SSD 2000-31
12/2000

Technical Attachment

Development of a Deployable Impromptu Mesoscale Web-Enabled Network

Pat Welsh
SOO, NWS Forecast Office
Jacksonville, Florida

Over the past summer staff members at WFO Jacksonville collaborated with researchers and students at the University of North Florida's Applied Global Research Laboratory to develop a COMET Partners Proposal. COMET support was obtained to explore development of a redundant and "impromptu" data network using the Jini technology. Jini is a part of Sun Microsystems Java technology, familiar to many of us because of its Web applications. The Jini technology allows a microprocessor with embedded Java to spontaneously announce its presence and "discover" any other Jini-equipped sites within its radio range, and then auto-connect to an existing packet radio network (called a "join"), or even to start a TCP/IP packet radio network with the first partner site it "discovers." The microprocessors and sensors then automatically build a network of "store and forward" network sites and report their data to a master station connected to the Web. The webserver (remember, this is a microchip on a 3" by 5" circuit board) then puts all the data on the Internet. Why is this important? While this seems (and is) rather vague and abstract, the concept has several applications in terms of a number of meteorological sensors which could comprise a local mesoscale observing network. We have now completed a demonstration of a three-station "proof-of-concept" system, and the following is a description of that effort and some of its potential NWS applications.

First, consider placing a ring of inexpensive instruments around a wildfire and having them automatically establish their own network and report their data into the IMET's laptop, providing constant updates of temperature, humidity and winds around the wildfire. The IMET laptop could then forward the data via cell modem to the LDAD at the WFO, which in turn could then post it to the Web for access by the emergency management community almost simultaneously.

Second, a relatively huge cost associated any mesonet is communications, but if an existing site has communications (landline or satellite DCP), then we could create a network (web) of radio reporting sites extending from that sensor outward to two to three times the radio range. Thus we could establish around thirty reporting stations communicating through a single site's connection (the theoretical limit is hundreds of sensors).

Third, should any sensor or sensors fail, the network will spontaneously reconfigure itself to continue reporting (via store and forward) of all sensors that can establish any path to the new server station.

Three UNF students have been working with WFO Jacksonville personnel on this COMET project. In addition to reciprocal visits and tours, they have been actively discussing each aspect of the project. John Sarman (our hardware guru) just graduated with a degree in electrical engineering, and he was the impetus for initiating the project last summer. Geoff Begle (JAVA and programming guru) just graduated with a degree in computer science; and Chandra Olsen, a Masters student in computer science, is the project coordinator, webmaster, theoretician, and general clean-up hitter.

On November 17, 2000, John, Geoff and Chandra conducted an operational demonstration of the project in which three sensor packages were turned on and initiated the network without human intervention; then automatically the sensors reported their wind speed, direction and temperature to the Internet via a simple GUI consisting of a thermometer object and speed and direction gauge objects. While this was a simple demonstration using very short-range radios, it successfully demonstrated the impromptu networking, including the discover and join using Jini, and the passing of meteorological data to the Internet. Implications of this successful test were immediately obvious to observers who perceived applications for expanding existing hydrologic observation networks, for critical cold air outbreaks during the winter, and similar operational situations.

The next stage in this project is to obtain grant money to deploy in the field a set of sensors, and

to "harden" and improve the packages with more powerful radios, solar panels and battery packs. Future research will focus on using a GIS webserver to autonomously create updated maps of the data every few minutes and post those to the Web.

Reference links...

Sun's JINI information (very technical): http://www.sun.com/jini/

UNF Applied Global Systems Lab: http://europa.cocse.unf.edu/

The students working on this project have identified themselves as the Jimini Research and Development Group, specializing in Jini wireless applications. Their Web site is at http://www.cise.ufl.edu/~cso/jini/. Chandra Olsen will soon update the group's Web page to include recent events described above...as soon as she defends her Master's thesis, which deals with the more theoretical underpinnings of the project. (Chandra is a student at the University of Florida, but she joined the UNF project when she learned how it was an application of her thesis work. Her contribution has been as a valuable - and unfunded - participant in the project.