Section 1: Project Objectives and Accomplishments
Recent work with high resolution numerical forecast models has shown that these models can be used as a diagnostic and forecast aid in assessment of high winds and rainfall amounts at very high mesoscale resolution (~4 km scale) compared to national models with synoptic resolution (~40 km scale and higher). Information of this quality and scale would provide Emergency Management officials with improved forecasts and in severe weather events such as hurricanes, could also provide initial assessments of damage. However, the development of this capability is currently limited by gaps in communications and computational capability between agencies. Among the existing shortfalls for creating high resolution modeling at such scales is the lack of the communications capability to link existing data sources (and deployable sensors) to the modeling agencies.
Recent advances and developments in computer microprocessor electronics, networked radio and packet switching communication technologies offer new opportunities for developing innovative solutions to this persistent problem, without incurring the burden of recurring costs associated with existing communications methods.
The general goal of our research is to develop a prototype weather station network that employs a new concept called 'Impromptu' Wireless Network technology. In our concept system, each field unit in the system can acquire weather data for its region and return the results to a centralized base station that can then relay the real-time station data to the Internet. The sensor data can then be viewed and the sensors themselves can be controlled using a Java-based web-browser interface. The primary objective of this research project was to develop and evaluate the 'impromptu' network component of this concept using Jini technology.
On 17 November 2000, a "proof of concept" demonstration was successfully conducted at the University of North Florida Applied Global Systems Lab. During this test, a wireless impromptu network of weather sensors was autonomously established without human intervention, and the data was automatically posted on the Internet web site. Witnessing this event were representatives of the National Weather Service, the Florida Division of Forestry, the U.S. Navy, the Florida Division of Emergency Management, and University researchers. Since that first demonstration, continued development of the system has occurred and has been presented at several conferences (listed below).
* Developed a system that uses packet-switching radio 'wireless Internet'
communications technology for data transfer.
* Developed a system that anticipated that individual units would fail, or be destroyed, and allowed for continued operation of the remaining network components in the absence of any specific individual component or station, by automatically reconfiguring the network topology.
* Developed a system that provided real-time Internet viewing of weather data from the field sensors (see: http://188.8.131.52/)
* Developed a design that utilizes recent advances in embedded microprocessor technologies to provide inexpensive, low power-consumption computerized control and data processing for each weather station unit.
* Developed a system design that will allow for rapid and efficient deployment of units in the field so that researchers can conduct impromptu study of specific weather events.
* Developed a system design in which the individual weather station units will be relatively inexpensive so that many units can be deployed.
* Developed a system design that could minimize field maintenance, communications limitations, and power requirements.
The impromptu network approach is a potentially revolutionary approach for managing networks of field data sensors since it implies that they can be very quickly and easily deployed in the field with minimal setup required. Since the design of the network is 'impromptu', the number of field unit devices in the network can shrink and grow with little (if any) impact on the overall operation of the network system. New field unit devices can be added to the system without the new device requiring a priori knowledge of the existing network structure. It is important to note that there is a requirement that each field unit device be in radio communications range of another field unit device that also exists on the network.
Section 2: Summary Of University/ NWS Exchanges
The Applied Global Systems Lab at the University of North Florida worked hand-in-hand with Dr. Pat Welsh of the NWS Jacksonville office. Hydrologist Joel Lanier from the Tallahassee NWS office attended one of the early meetings with Pat Welsh and agreed that there were also hydrologic uses for such a data network. Meetings with the project team were scheduled every two weeks to review progress, generally alternating between UNF and the NWS JAX office. Team members were invited to NWS JAX to observe operations and to discuss the office networks and server configurations with the NWS JAX Electronic Systems Analyst, Mr. Art Wildman and his assistant, Shane Still. In addition they discussed future careers in the NWS as Information Technology specialists. Other topics included the use of multiple communications technologies and operating systems including Windows variants, several flavors of UNIX and Linux. These operating systems were discussed with the strengths and weaknesses of each from an operational perspective. The team members talked with the NWS JAX forecasters on duty, and how they used computer systems to facilitate forecasting.
In the biweekly meetings, Dr. Welsh was personally involved in developing team goals and assisting the team in making decisions about how to proceed in the development of the network software. Dr. Lambert was involved in nearly every aspect of the prototype development and provided frequent guidance to the team.
Both principals provided information on how to present this project and how to direct the effort toward current weaknesses in how meteorological data is obtained and communicated. The possibility of exploring future cooperative proposals was discussed.
Several meetings were held between Dr. Welsh and University of North Florida administrators including the Dr. Jim Collom, Director of Sponsored Research, Dr. Neal Coulter, Dean, College of Computing Sciences and Engineering(COCSE) , Dr. Jerry Merckel, Assoc. Dean, COCSE, Dr. Joe Campbell, Director, Division of Engineering. The result of these meetings was a tangible plan for establishing a long-term relationship between UNF and NWS Jacksonville.
Section 3: Presentations and Publications
The graduate students participating in this project presented the results in a presentation to a joint session at the annual meeting of the Central Florida Prescribed Fire Council during October 2000 in Tampa, Florida. They also presented a live demonstration of the technology during the poster session. Over 150 firefighters and disaster managers attended the joint session and about 60 attendees observed the live demonstration. Numerous firefighters and the NWS Ruskin (Tampa) SOO, Charlie Paxton, became interested in the project. He has been following the effort by email.
A "proof of concept" demonstration and evaluation event was conducted for an invited audience of researchers, representatives from Florida state agencies, DOD (Navy) representatives, and Jacksonville NWS at the UNF Applied Global Systems Laboratory on Friday November 17, 2000.
Dr. Pat Welsh and Dr. David Lambert were invited to make a presentation to the joint session of the U.S. Weather Research Program Science Symposium and the Interdepartmental Hurricane Conference during March 2001 in Orlando, FL. Approximately 100 attendees were exposed to this networking technology and data communication concept. Interest was expressed by US DOT (Shelley Roe) and Florida state EOC (Andy Devanas) as well as NWSHQ staff (Tim Ross) and NOAA ETL (Dr. Albin Gasiewski). Most of the Florida NWS offices were represented at the conference. (Post-print of the conference proceedings is in press.)
John Sarman, one of the graduate students working on the project, participated in the "invitation only" international Jini™ Network Technology Workshop held during April 2001 in Princeton, NJ. The workshop, which was sponsored by Sun Microsystems, Inc., brought together the most influential Jini network technology researchers (including the inventors) and industry representatives from around the world. Mr. Sarman was the only student invited to this "insider" conference. He presented the results of this project at the workshop and gained invaluable insight and advice about how to proceed with development of this application of the technology.
Section 4: Summary Of Benefits And Problems Encountered
Benefits to the University
In addition to the new familiarity between the two groups, we have developed an appreciation of where we may help one another in the future. The project has also provided an excellent opportunity for students at the Applied Global Systems Laboratory to create prototype systems that serve a definite purpose. Both Computer Science and Electrical Engineering majors have shared in the findings and related these to their coursework. The student interactions with forecast staff and the electronic systems analyst (ESA), Mr. Art Wildman, at the NWS Jacksonville office has provided real world experience for the students.
Collaboration was initiated between the NWS office and the University of North Florida Applied Global Systems Lab in the development of a local ecosystem research community. This collaboration will also include participation by the nearby Estuarine Research Reserve, Jacksonville University, and the Timucuan National Preserve.
Finally, this project formed the basis for the development of a follow-up COMET Cooperative Research proposal that has been submitted to UCAR. The goal of the Cooperative Program proposal is to continue development of this concept with the end product being an operational mesoscale weather station network for Northeast Florida that will utilize the impromptu network technology developed in this Partners Project.
Benefits to NWS
This research indicates that Jini technology can be used successfully to create "impromptu" wireless networks that can enhance the retrieval of meteorological data.
The project has lead to new and useful insights into this technology's ability to provide meteorological data at mesoscale spacing without the huge recurring costs of conventional communications.
This project has allowed for the "force multiplier" concept which would allow multiple weather stations to be deployed in an area to increase the number of reporting sensors in a mesoscale array, thus enhancing the density and accuracy of weather information around the data site. This is particularly true in cases where there is a lack of existing weather stations due to limited communication range.
Additionally, our research is leading us to develop concepts and techniques that would allow for rapid deployment of a network of meteorological sensors during special weather events. Such an array would also be useful for managing wildfires, hazardous material accidents, and severe weather, including hurricane landfalls.
Several obstacles were encountered
* The original microcontroller proposed did not contain sufficient memory
capacity and another microcontroller was selected.
* Windows encounters numerous difficulties in programming and running Jini technology, consequently, the future direction of our research will move towards using embedded Linux systems.
* The initial setup of Jini was more complex than anticipated due to a lack of helper facilities and appropriate APIs. Once again, the move towards Linux-based systems may remove this obstacle.
* The reed switches on the weather stations that indicate direction were too fragile and could be broken too easily during weather station assembly. This problem was unique to the inexpensive sensors used for the prototype system. Future prototypes will incorporate much more robust hardware.