top lines Ernest Orlando Lawrence Berkeley National Laboratory
Computing Sciences
Organization Chart
Research Division
NERSC Center Division
Privacy & Security Notice
Questions & Comments
Berkeley Lab A-Z Index Phone Book Search
Computing Sciences' Deb Agarwal to Help Develop Test Ban Treaty Monitoring System

April 9, 1999

Verifying compliance with the Comprehensive Test Ban Treaty presents a host of technical and political challenges.

While the political issues among the more than 150 nations signing the treaty are being handled in the diplomatic arena, Berkeley Lab's Deb Agarwal is lending her technical expertise to helping overcome the technological hurdles.
Deb Agarwal

Since taking on the assignment earlier this year, Agarwal of the Information and Computing Sciences Division has traveled around the country meeting with experts in the field of sensor data collection, storage, transmission and authentication. The next stop on her itinerary is Vienna, headquarters of the Comprehensive Test Ban Treaty Organization, where she will be staying for the next three months.

Agarwal's immediate task there will be to serve as an independent expert and to help the organization determine if "reliable multicast" technology is a cost-effective way of handling the large amount of data involved in verifying that no nation detonates a nuclear device. If the technology is deemed appropriate, Agarwal will then help specify a communications protocol for the international monitoring system.

She wrote her Ph.D. thesis on reliable multicasting and has been providing reliable multicast service to the DOE Collaboratories for the past four years, but assessing whether it could work effectively on a global scale for verification of the treaty is a significantly different problem.

"The more I become involved, the more I learn," she says. "I could end up having a significant impact on how the communications system is built to verify that no one conducts another nuclear test. That's cool!"

IP Mulitcasting, or the transmission of a single data stream to several different sites can provide an efficient mechanism for transmitting data. It is, however, based on the "best effort" concept and data gets lost - which is unacceptable for treaty verification. "Reliable multicasting" provides a layer of reliability by retransmitting when data is lost, and by having systems in place for avoiding network congestion.

Although a number of federal agencies are contributing to the technology needed for treaty verification, they also realized that multicast expertise was needed.The Lab's Stu Loken and Bill Johnston were contacted, and they recommended Agarwal.

"Deb has brought tremendous energy and expertise to the collaboratory program and especially to the multicast area," said Loken, director of the Information and Computing Sciences Division. "It's great to see her work being recognized in this important international project. What she learns in this new effort will be of great benefit to the collaboratory and Strategic Simulation Initiative projects as well."

The challenge in building the international monitoring system is to link some 337 monitoring stations around the globe. These include arrays of sensors designed to measure and transmit data about seismic, hydroacoustic and infrasound events and evidence of radionuclides. Although all the data will be transmitted, stored, analyzed and disseminated from the International Data Center in Vienna, some nations have asked that they also receive all the raw data simultaneously. Reliable multicasting, Agarwal says, has the potential to meet both needs, as well as circumvent the possibility of having the entire system crash if the Vienna center fails.

The global array of sensors in the atmosphere, oceans and underground will detect a wide range of activities, such as nuclear blasts, earthquakes, lightning, meteors hitting earth and mining explosions. Data from these events, which are expected to number from 100 to 500 per day, will then be analyzed to determine the time, location and probable cause. The treaty organization will issue bulletins containing both raw data and analyzed data. If a suspicious event cannot be resolved through consultation and clarification, each nation signing the treaty has the right to request an on-site inspection.

Various United States organizations are helping to pull all of this information together. Sandia, Los Alamos, Pacific Northwest and Lawrence Livermore national laboratories are developing sensors and data analysis software. The Air Force Technical Applications Center, which has operated a global nuclear event detection system since the first bilateral treaties were signed in the 1960s, is involved, as are the Defense Threat Reduction Agency, the Center for Monitoring Research and private sector firms. Agarwal's work is being funded by the CTBT Research and Development Program in DOE's Office of Nonproliferation and National Security.

Agarwal has visited many of the agencies, she says, and the importance of her role has been shown by the fact that top managers are willing to clear their calendars and meet with her for an entire day or more. "Everyone involved has been looking for answers to their questions and one of the things I've been able to contribute already is helping to find out those answers," she says.

Some of the questions needing to be answered include system-level issues regarding how the communication protocol pieces fit together, how to distribute the data to multiple subscribers so they all receive it simultaneously, and how to ensure that the data is reliable. The CTBTO is an international organization and each nation has its own set of data requirements. Some nations, such as the United States, have chosen to receive data from their territory before it is distributed to the rest of the world, and some nations want to receive the raw data from some or all of the sensors in the entire network. Some other nations have asked to receive the data from their territory simultaneously with its transmission to Vienna. The communications system needs to allow all of these requirements to be satisfied. Compounding the issue is the requirement that all data must arrive in Vienna within five to 10 seconds of its creation.

For Agarwal, who gave up a job doing off-line programming of assembly line robots for General Motors to earn her Ph.D., an opportunity such as this is one of the reasons she chose to work at a national lab.

"When you go to work for a company, there's a limited scope of what you could be involved in," she says. "At the national labs, you have a lot more opportunity to be involved in a broader space. Working on this project is very exciting."

For more information about the program, visit:

  U.S. Department of Energy · Office of Science · SciDAC · ASCR · University of California Science logos ASCR logo SciDAC logo Office of Science logo DOE logo UC seal