TECHNOLOGY VS. TERROR
After the Sept. 11 attacks, investigators followed a trail that led from the ashes of the World Trade Center and the damaged Pentagon back across the country to airports, flight schools, apartments, computers, personal possessions and dozens of other bits and pieces of information. Eventually, investigators figured out who the hijackers were and how they carried off their attack.
Investigations following last year’s train station bombings in Madrid and this year’s underground bombings in London also moved swiftly to find the people responsible.
The clues that enable investigators to explain events after they occur are also available beforehand. Is it possible to recognize telling clues before a terrorist event? How can analysts determine which clues are important and which are not? How can they weave appropriate information into patterns that will make it possible to detect and stop a terrorist attack before it happens?
That’s the goal of a new scientific discipline called visual analytics being developed by the Department of Energy’s Pacific Northwest National Laboratory (PNNL) in Richland, Wash.
PNNL and the rest of America’s national laboratories are taking on many of the largest, most difficult technological challenges in the war against terror. Is it possible to predict when, where and how the next terrorist attack will occur? Is it possible to detect the silent, invisible and deadly release of a chemical or biological weapon before people begin to get sick? How can first responders take advantage of the increased situational awareness that emerging technology will soon make available?
Out on the intellectual and technological frontier of the war against terrorism, the nation’s national laboratories are investigating these questions.
VISUALIZING THE UNIMAGINABLE
Analysts often talk about chatter that indicates the likelihood of a terrorist attack in one or another region of the world over a period of time.
Chatter is an amorphous collection of information obtained in legal and sometimes perhaps not-so-legal ways — from newspapers, Internet sites, government intelligence efforts, e-mail, telephone conversations and millions of other reliable and unreliable sources. Analysts read and evaluate chatter; then, they draw conclusions that may or may not be prescient.
Visual analytics is an emerging scientific discipline that uses the information-processing power of computers to help read and evaluate chatter.
Visual analytics systems employ mathematical formulas called algorithms to count up mentions of names, locations, dates, times, weapons and a million other details that spill out of chatter. These systems then organize and re-organize this information into patterns presented in the form of graphical charts. By processing data in different ways and continually updating the results with new chatter, scientists believe the technology can identify patterns that may help governments get out in front of terrorist planning.
“We have found that humans are pretty good at creating data structures and organizing data to discover something that they think might be there,” says Jim Thomas, director of the Department of Homeland Security’s National Visualization and Analytics Center at PNNL. “What we are not good at discovering is what we have not perceived to be a possibility ahead of time.
“The questions can be confusing,” Thomas says. “But in order to discover the unexpected, we need to ask what is going on, and what is not going on. Then we probe the information space for patterns related to these questions. Upon finding patterns, we can look for supporting logic.”
Clues suggesting that terrorists were planning to attack the World Trade Center with hijacked commercial airliners existed before Sept. 11, 2001. For at least two reasons, no one tied those clues together. First, few even imagined that what eventually happened was even possible. Second, while the important information about Sept. 11 was available, masses of other information obscured its importance.
The hope of visual analytics is that power computers capable of mathematically processing enormous amounts of data might next time find patterns worth investigating — not necessarily patterns leading in a straight line to a terrorist plot, but patterns interwoven with investigative opportunities that might have been worth pursuing.
Predicting the future is, of course, uncertain. But what if visual analytics evaluations could suggest, for example, that current information patterns related to biological and chemical weapons are more powerful than patterns suggesting radiological attacks. Wouldn’t that be an important discovery?
What gives this admittedly unusual idea credibility is growing computer-processing power. Right now, a visual analytics system can evaluate 500,000 documents. “Over the next three to five years, we will begin to build suites of technologies that will enable analysts to look at information spaces containing as many as a billion items,” Thomas says. “A computer analysis of that many pieces of information would enable analysts to develop hypotheses based on patterns built with enormous amounts of supporting data. We will be able to discover rapidly what is and isn’t in a particular information space.”
AIRBORNE BIOLOGICAL THREATS
Late last year, the Lawrence Livermore National Laboratory in Livermore, Calif., introduced a device that sniffs the air and tests for more than 95 possible airborne biological threats, including anthrax, plague, and a range of bacteria, viruses and protein toxins.
Called an Autonomous Pathogen Detection System (APDS) the lectern-sized device can be placed at airports, office buildings, performing arts centers, mass transit systems, sporting arenas and anywhere else a biological attack might be launched.
The APDS innovation is that the machine works by itself. It sniffs the air and collects samples. It runs two tests and draws two independent conclusions, thus ensuring against false positives. Within two hours, the system can collect a sample and deliver results over the Internet to a monitoring station. “After hundreds of thousands of tests, we have had no false positives, says John Dzenitis, a chemical engineer and project leader on the Livermore APDS project.
According to Dzenitis, the APDS is the latest version of a technology originally called the Biological Aerosol Sentry and Information System (BASIS), a manual device that required operators to collect filters and deliver them to a lab for testing. Eventually BASIS technology was incorporated into the BioWatch program now managed by the Department of Homeland Security.
APDS improves on the manual BASIS and BioWatch by automating the tests and slashing the time required to collect and test samples from days to two hours.
The next step for this technology will be to produce versions that are even smaller, faster and less expensive. “Right now, we are field testing a project called Bio-Briefcase, which is smaller and less expensive than APDS,” Dzenitis says. “In the same technological pipeline, we are also developing a biological aerosol mass spectrometer (BAMS) that will take another step forward by speeding the tests.
“The goal is to make sure that we can deliver actionable information to public health people — in time for them to make informed decisions about how to respond to an airborne biological attack,” he adds.
SENSORNET
The Oak Ridge National Laboratory in Oak Ridge, Tenn., is integrating sophisticated chemical, explosives, nuclear and radiological sensors into situational awareness networks.
“Our goal is to develop sensors and sensor systems that can communicate across networks,” says Frank A. DeNap, the SensorNet program manager. “When a sensor is added to a network, it should be able to advertise its presence and say, for example, that it is a radiation detector, authorized to join this network, at this location. These sensors will also be able to disseminate information to monitoring stations staffed by police, fire and emergency medical first responders.”
Network sensors will be mobile, DeNap continues. Sophisticated sensors are too expensive for broad distribution. Instead, they will be set up as needs arise in high-threat areas.
Currently, the Oak Ridge SensorNet program is operating in a half dozen mobile test beds. Pilot SensorNets are located at the Port of Memphis and Ft. Bragg, N.C., for example.
At the Port of Memphis, a shipping transfer facility dealing with hazardous chemical shipments has deployed a SensorNet to report chemical releases, deliver information to a site that models plumes in light of weather conditions during releases and follows events with camera surveillance. “If something happens, the emergency management authorities will receive information about the affected area and be able to take steps to mitigate the consequences of the event,” DeNap says.
At Fort Bragg, N.C., a SensorNet goes even farther in delivering a comprehensive picture of the security of the installation. Fort Bragg has installed chemical, biological, radiation, explosive, intrusion and fire sensors in a single network. Should something happen, a sensor itself will call 9-1-1, specify the problem and even deliver schematics detailing the location of the problem.
WHAT HAPPENS WHEN THE ALARMS GO OFF?
When any kind of security sensor goes off, who investigates? The fire department? Police department? Emergency-medical technicians? What recommendations will the investigating agency make? Should people in the area stay inside? Should they exit the building? Once outside, should they wait for first responders to help them? Should they go home? What emergency units will the investigating agency summon?
If there is a biological event, will one or another agency offer antibiotics? If so, how? Will it be necessary to quarantine people or evacuate people? Should the heating, ventilating and air conditioning system be turned off or turned up?
These questions are relatively easy to answer for one facility. But what if the questions affect the New York metropolitan region or Southern California?
“You have to collect and present information in a way that facilitates human decision-making at many different first-responder levels,” says Duane Lindner, program manager for chem-bio national security at the Livermore, Calif., Sandia National Laboratory, a sibling of the Albuquerque, N.M., Sandia Lab.
A Sandia program studies the problem of planning appropriate chains of responses to alarms raised by regional chemical, biological, radioactive, nuclear, explosive and other security technologies tied into alarm systems.
Sandia also studies the problems related to returning a facility to service after a major event. The Hart Office Building in Washington, D.C. closed for more than a year after an anthrax attack. If a biological attack shut down an international airport for a year, the economic costs would total in the millions of dollars per hour for every day of the shutdown.
“We have to learn to make facilities operational again as fast as possible,” Lindner says. “The protocol includes carrying out a contamination assessment, assessing appropriate decontamination approaches, getting procedures approved and sampling the environment when the decontamination process concludes.”
As the war against terror comes into sharper focus, so does the value of imaginative thinking about security technologies. By designing software to help analysts track terrorists before they strike, technology is attempting to peer into the future and ward off attacks. By engineering better, faster, cheaper sensors and sensor networks, technology is helping to direct and manage the human response to terrorist events. By studying and refining the interaction of people with security technology, researchers are speeding and improving society’s recovery capabilities.