More Eyes in The Sky May Not Generate Better Intelligence
Under orders from Defense Secretary Robert Gates, the military services are moving to accelerate the production and deployment of surveillance aircraft.
This mad rush to deploy more spy planes over Iraq and Afghanistan is meant to help U.S. troops hunt down their shadowy enemies.
But more eyes in the sky may not necessarily translate into better information on the ground, according to military officials and scientists.
The problem is that no matter how many more dozens or hundreds of manned and unmanned aircraft hover over war zones, the data that they provide have limited value because they cannot be shared across the force.
Another impediment is that the massive amounts of data being collected cannot be analyzed quickly enough to be useful.
Incompatibility between Air Force and Army databases often hinders the flow of information in Iraq and Afghanistan, officials said. Streaming video from the Air Force’s unmanned aerial vehicles, or UAVs, is not always available to Army units in real time, nor is the Army’s UAV data automatically made available to Air Force commanders. As a result of this lack of interoperability, much of the intelligence that UAVs collect cannot be acted upon because it’s not accessible in real time.
“The key issue is not the aircraft, it’s getting a common picture,” said Brig. Gen. Stephen Mundt, director of Army aviation. “The debate needs to be about the ability to see all things in the air on a common picture” rather than on how many aircraft should be deployed, he told reporters.
Air Force officials have argued for some time that this situation could be solved by turning over control of all Defense Department medium- and high-altitude surveillance UAVs to the Air Force, but the Army has resisted, claiming that its commanders want to be in charge of their own UAVs rather than have to work through the Air Force’s bureaucracy.
The inter-service spat over UAVs has gotten so ugly that both services have given up trying to resolve it on their own and now expect the dispute to be settled as part of a broad Pentagon-wide review of military “roles and missions.”
Mundt contends that no matter who owns the UAVs, everyone should still share a common picture of the airspace, and that the services should have a single “blue force tracker” for both manned and unmanned aircraft.
“That’s not something we thought about before the war,” said Mundt. During the past several years, both services have tried to come up with fixes but have not succeeded, he said.
“I should be able to see how many UAVs are in the area, where their cameras are pointed,” said Mundt.
Both Army and Air Force commanders should be able to identify which aircraft are available in the area and request overhead surveillance from the UAVs that are closer to them. The streaming video also should be piped into a single database that can be accessed by individual soldiers equipped with laptop computers and by commanders at headquarters.
“It’s about sharing information horizontally,” Mundt said. “It’s a Defense Department problem.”
All services are to blame for this, he said. Each service builds its own UAVs and sensors, all from different vendors. “You end up with stovepipe information channels,” which means the data have to be sorted out and disseminated, thus causing further delays.
“We have to break the gridlock on this,” Mundt said. “When every OEM [original equipment manufacturer] builds something slightly different, it doesn’t help anybody.”
Gates’ latest mandate directs the services to push more intelligence, surveillance and reconnaissance collectors to the battlefield relatively quickly. Most of those ISR systems will be unable to share data, Mundt said. “We are going to do it the manual way we’ve been doing — collect, move, fat finger, swivel.”
The Army and the Air Force have been working to bridge the interoperability gaps for the past couple of years, he said. “But it doesn’t get solved overnight.”
The Defense Department’s uncoordinated approach to developing and deploying ISR systems was the subject of a March 2008 Government Accountability Office report.
During the next seven years, GAO estimated, the Defense Department plans to spend more than $28 billion to develop, procure, and modify 20 major airborne ISR systems. Of concern to GAO is that the Pentagon does not manage those assets efficiently. The 2004 National Defense Authorization Act stated that the Defense Department would attempt to fully integrate ISR capabilities and coordinate the various programs of the services, intelligence agencies, and combatant commands.
“The Defense Department has initiatives underway to improve the integration of its ISR investments,” GAO said. “However, the department lacks key management tools needed to ensure that ISR investments reflect Defense Department-wide priorities.”
The military’s ability to gather useful intelligence from ISR systems also faces a major obstacle: a data glut. Analysts are not currently equipped to cope with gobs of data and parse them in real time. This problem only will worsen as more surveillance aircraft are deployed.
The Defense Department has a tough time “managing the prodigious quantities of data produced by new sensor systems increasingly being planned and deployed,” said Roy F. Schwitters, a professor of physics at the University of Texas at Austin and chairman of the JASON Defense Advisory Group.
JASON is an independent group of scientists that advises the U.S. government on matters of science and technology.
The value of ISR technologies currently is being “hampered by inadequate analysis tools which are not suited to handling the large quantities of data created by the systems,” Schwitters told the House Armed Services subcommittee on terrorist and unconventional threats.
The complications caused by a data glut are especially acute in Iraq as a result of efforts to detect roadside bombs and insurgents who plant them — all of which requires vast amounts of streaming video and high-resolution imagery.
A single aerial reconnaissance system may use 100-megapixel cameras operating several frames a second, said Schwitters. Those sensors alone can generate 10 to 100 terabytes of data, he noted. “These systems generate tens to hundreds of terabytes in a day of observation.”
The bottlenecks are not necessarily the result of a bandwidth crunch, Schwitters said. They are caused by delays in examining the data. “It’s the analysis that must keep up with the flow of data to avoid pile-up,” he said.
Schwitters said the situation is reminiscent of the “hilarious TV episode of ‘I Love Lucy’ where Lucy and Ethel are at the chocolate factory and the chocolate just gets out of control, and you never get back in gear.”
The human mind can’t keep up with that much information, he said. “So we need new ways to handle this data, and that’s the issue in front of us.” The traditional approaches of compressing video files actually can harm the analysis and therefore the value of the imagery, Schwitters said. “You lose critical information that cannot be retrieved unless that data are handled properly. These are simply new things that we need to deal with in addition to just managing the volume of data.”
Allan Shaffer, principal deputy director of defense research and engineering at the Defense Department, testified at the same hearing that the current overabundance of information is just the beginning. “The current projections are for the data volume of the defense systems to grow by as much as a factor of a billion over the coming decade,” he said. “The defense science and technology community is already planning for this growth.”
The Pentagon set up a “data-handling technology focus team” in fall 2007, said Shaffer. The team concluded that the issue is not just the size and amount of information, but the time that is needed to act on it, he said. “The team recommended several actions, from revamped architectures to processing closer to the sensor.”
Under a project called “intelligent agent technology,” the Army is probing ways to sift through data faster by relying on artificial intelligence systems, said Thomas H. Killion, chief scientist and Army deputy assistant secretary for research and technology.
An intelligent agent is a piece of software that looks for specific attributes in the data and can prompt a human operator when it finds items that it was programmed to identify. “It’s hard to have an analyst look at a strip map from the day before and the one that you just took an hour ago and compare and find all the little changes that may have occurred, whereas the computer can say, there’s a change here, here and here,” said Killion. “He can look at those and he doesn’t have to spend 12 hours pouring through that strip map; he can do it in 10 minutes perhaps.”
Intelligent agents, however, only go so far, Killion conceded. “I went to graduate school in an era when they were talking about artificial intelligence computers that were going to be just as intelligent as people any day now, you know, and unfortunately, that was quite some time ago. They’re not there yet.”
Please email your comments to SErwin@ndia.org