Collaborative tech morphs data into intelligence

To more quickly and effectively make battlefield decisions, warfighters need the final product of shared knowledge and collaboration: actionable intelligence.

Defense and intelligence community researchers and the intelligence, surveillance and reconnaissance industry are looking for technologies that help intelligence analysts and warfighters share information more easily and meaningfully. Rather than merely pushing sensor data to troops, the goal is to create a common operating picture that facilitates faster decisions in the field.

Analysts say there have been great strides in ISR data collection during the past decade. Systems such as the Distributed Common Ground System (DCGS) Integration Backbone (DIB) have provided the foundation for a common architecture that enables more data sharing throughout the Defense Department and intelligence community.

“Not an awful lot of technology investment needs to be made in this industry to make significant progress beyond where we are,” said McClellan “Guy” DuBois, vice president of Raytheon's Intelligence and Information Systems business unit and chief executive officer of Raytheon UTD. “No miracle has to happen to make this happen. What needs to be done is more focused on the overall architecture itself.”

“What started out almost a decade ago as an Air Force program has now broadened out to the DIB being an accepted standard” by the Office of the Secretary of Defense, DuBois said. “And it's moving a lot of information around both vertically and horizontally. The growth of applications on the DIB has made information sharing much easier. It hasn't solved all the problems, but by getting it down to the lowest common denominator, it’s made it a lot easier to innovate.”

For example, Raytheon has developed a multimedia intelligence application, called the Raytheon Android Tactical System (RATS), that uses Google’s Android mobile phone operating system to deliver imagery and full-motion video to troops in the field. “With RATS, we’re taking a mobile phone and turning it into a collection and dissemination tool. A soldier in the field can, for example, see someone like Osama bin Laden and take his picture and pump it back into the system through the DCGS Backbone.”

Although sharing within DCGS is becoming easier, problems arise when trying to move intelligence information across domains — from one classification level to another, from the intelligence community to DOD, from U.S. systems to allies', or from DOD agencies to homeland security agencies. Another problem is identifying intelligence that analysts need to send to warfighters on the front lines while moving forward to analyze the new data that warfighters send.

“At one level, there's this information management problem,” said William Regli, professor and director of the A.J. Drexel Institute for Applied Communications and Information Networking at Drexel University. “There's an overwhelming amount of data coming in, and somehow all of it has to be filtered and rarefied. And people have to make decisions based on that data.”

Working Across Domains

Search and signal-processing technologies can help resolve those problems by capturing information from multiple data streams without the work of an analyst, Regli said. But, he added, they don’t address the second problem: "How do you manage hypotheses, theories [and] do the diagnostics on top of that data?"

That calls for collaboration across domains to tie together the analysis of related data so the analysis can be more effectively and efficiently completed, consolidated and shared with people who need the results.

There are a number of ways that analysts can share intelligence data across domains. Many of those systems are on the Unified Cross Domain Management Office’s Cross Domain Baseline list, and they are systems for intelligence information sharing that have been deployed, tested and approved for wide use by the office.

Access solutions are one category of systems on the baseline list, said Ed Hammersla, chief operating officer of Trusted Computer Solutions, a vendor of three approved mechanisms for information sharing.

“Access solutions allow you to gain access to multiple networks from a single device," Hammersla said. "But it doesn't allow the transfer of information between them.”

That allows analysts to work with information from, for example, the Joint Worldwide Intelligence Communications System network — which contains information classified up to top secret and sensitive compartmented information — at the same time as they access systems on the Secret IP Router Network and Unclassified but Sensitive IP Router Network. Trusted Computer Solutions’ access solution, called the Trusted Thin Client, was deployed as part of the new combined air operations center for Central Command in October 2009. Hammersla said the system uses a single server that connects to all three networks to serve thin-client systems via a single network, allowing the Air Force’s 350th Electronic Systems Group to set up the combined air operations center with fewer desktop computers and less administrative overhead.

"We added far more robust systems, all while enhancing the interoperability of each work area and reducing space and power needs,” Air Force Capt. Dennis Smith, the 350th Electronic Systems Group fielding flight commander, said in an Air Force statement. “We also expect this to be more efficient because it will reduce the power requirement for the computers and the air conditioning.”

However, access solutions don’t solve the collaboration problem because they don’t allow multiple-level security access and cross-agency data sharing. That’s where guard systems, as Hammersla calls them, come in. Those systems sit at the boundaries between networks and provide ways for high classification information to be downgraded and shared with lower-classification networks. They also can pass information from lower-classification networks upward while preventing malware or attempts to hack into the classified network.

“Either of these — high to low, or low to high — can be automated or have a man in the loop,” Hammersla said. “There are automated guards to downgrade imagery. But there's also automatic low to high. A good example is Air Force weather data, which comes from unclassified sources but is needed on the classified networks for mission planning.”

Some guard systems require a person to approve the movement of information and complete the downgrade. “That's a producer/releaser construct, where one person creates it, another person approves its release,” Hammersla said.

But that sort of connection creates a problem in an information-saturated operation, Regli said. Although data crosses both ways, it’s hard for everyone to share what data they’re working on.

“There's too much going on,” Regli said. “So you can't have a human being sitting there at the firewall between networks looking at every intelligence hypothesis on either side and say, 'Hey, you people have the same idea. You should work together.'”

Text Mining

A variety of efforts have demonstrated promise in addressing that challenge. One example is the JWICS-based Analytic Space, or A-Space, a project of the Office of Analytic Transformation and Technology at the Office of the Director of National Intelligence. Although A-Space allows intelligence community analysts to collaborate on topics of mutual interest, it excludes everyone down the classification chain who could often benefit from the results of that analysis — or even contribute to the analysis.

Another approach is shared systems, such as the Intellipedia wiki created by the U.S. intelligence community. However, Intellipedia operates as three separate systems — one on JWICS, one on SIPRnet, and the Open Source Intelligence version on NIPRnet. There’s no effective way of moving data between the systems because the information is unstructured and does not have tags to ensure proper content transfers.

An effort by Accenture could help bridge the divide. Accenture is working on what Ryan La Salle, a director at Accenture’s defense industry group, calls a semantic wiki. "A wiki is unstructured information, and any structured data in the wiki is inconsistent," La Salle said. "A semantic wiki allows it to act more like a structured database.” With that structure, he said, it can be easier to move information across domains in an automated fashion.

Accenture is also working on ways to automatically populate a wiki with intelligence information. “Wikis require people to manually input data,” said Chris Zinner, also a director at Accenture’s defense industry group. “So it's one of the reasons we have what we have today [in failures to share information]. Whenever you ask someone to put information into a specific form, it slows them down. So we’re leveraging experts in our lab around text mining and looking at the question, ‘What if we used text mining technologies and tried to seed the wiki?’ We could pull in unstructured data, intelligence reports and automatically construct a wiki page.”

Text mining can’t generate a 100 percent solution, Zinner said. “But we can get it semi-automated. If we can leverage text mining to get 80 to 90 percent of the way, people can become data stewards rather than just entering data.”