Wednesday, August 01, 2007

Predicting IEDs?

The heroism and sacrifice of EOD technicians like Staff Sergeant Joshua Mattero (previous post) is a reminder of the herculean battle being waged in Iraq against the enemy's only viable weapons--Improvised Explosive Devices (IEDs) and their vehicle-borne counterparts, better known as VBIEDs.

By some accounts, the various task forces, working groups and agencies assigned to the counter-IED mission are spending upwards of $6 billion a year, and their budget will see further increases as the conflict continues. In defense circles, the IED/VBIED threat represent a rather grim "growth" industry, with DoD spending freely on almost any project or program that promises to diminish the danger on the ground.

In some respects, finding a solution for the IED problem is a bit like cancer research; in all likelihood, there will never be a single, "magic bullet" cure, but various programs and efforts can provide more options for mitigating the threat. The on-going troop surge clearly provides one of the best deterrents, eliminating terrorist sanctuaries where IEDs, suicide vests and VBIEDs can be manufactured, and distributed to the bombers.

Improved surveillance and better intelligence are also keys; as Michael Yon and others have noted, the terrorists littered the city of Baqubah with IEDs before the recent U.S. attack, but our forces largely avoided that threat, in part because Iraqi civilians pointed out where the bombs had been placed. U.S. Weapons Intelligence Teams, comprised of intelligence operatives, EOD techs and other specialists, have also proven effective in going after IED networks.

But in some respects, the progress has been agonizingly slow. Despite a dramatic reduction in U.S. combat deaths last month, IEDs remain the #1 killer of American troops in Iraq. Data from the Iraq Coalition Casualty Count database indicates that improvised explosive devices (of all types) accounted for 46 of the 66 combat fatalities in Iraq last month, or 70% of that total (You'll note that we do not include deaths from non-hostile causes in our totals, unlike the MSM).

Totals for April, May and June of this year are strikingly similar. IEDs were responsible for 66% of our combat deaths in April, 73% in May and 75% in June. And remember: our troops find (and neutralize) at least 40% of the bombs before they go off; without their valiant efforts, the toll of dead and wounded from IEDs would be significantly higher.

What's been missing from our counter-IED efforts (to date) has been a tool for predicting where bomb attacks are likely to occur. The problem isn't a lack of data; aerial surveillance, signals intelligence (SIGINT), human intelligence (HUMINT), operational reporting and other sources generate reams of information. But wading through that data has been difficult; as with other analytical tasks, it's a monumental challenge to simply separate the wheat from the chaff, let alone connect the dots.

But very soon, our forces in Iraq (and elsewhere) may get a powerful tool to perform predictive analysis of IED events. The Air Force's C2ISR (Command and Control; Intelligence, Surveillance and Reconnaissance) Battlelab, based at Langley AFB, Virginia, is working on a system called RADII (Rapid Assessment and Dissemination Information Infrastructure), which can be programmed to look for key data relationships, and dynamically generate actionable information from constantly-updated data.

RADII is being developed by Sensis, a defense firm that has traditionally specialized in air traffic control and air defense systems. A former colleague sent us an unclassified summary of RADII, which is now in its third development spiral, adding the predictive analysis capability. The information provided to us gives no indication of when the system will be operationally deployed; we can only assume that the generals and senior civilians running the counter-IED program are aware of RADII, and will put it in the field as soon as possible.

How is RADII different from other analytical tools? According to project managers, the system not only correlates information from a wide range of databases, it also employs a recursive learning algorithm that can be "trained" to discover emerging patterns in new data, and compare it to "historical" events of interest. In other words, if RADII finds a correlation between on-going events in a certain location (and finds a correlation to similar activities associated with past IED attacks), it will note the associate, and generate a warning for analysts, who can pass it on to units in the field.

RADII also allows the strongest correlations to be graphically displayed, in either a "links and nodes" wiring diagram, or on maps or imagery of the local area. The graphical depiction is particularly useful in showing how seemingly unrelated events and facilities may actually be part of an IED network, allowing more effective targeting of its elements, and the assessment of where future events are likely to occur.

But will it work? As with any analytical tool, RADII will be only as good as the information it processes--and the people who use it. False or inaccurate data will result in false alarms and wasted effort, and it's worth noting that Sensis and the Air Force are keeping the analyst squarely in the loop, allowing RADII to take advantage of user "refinements," based on their experience and expertise. Put more succinctly, the developers recognize that there's no substitute for a trained analyst, the RADII would be best employed with intel specialists who have past experience in Iraq and the IED mission.

Despite that, we believe that RADII could prove to be an important addition in the fight against IEDs. Sorting through vast databases, establishing correlations, comparing them to past bomb attacks and learning over time, RADII could finally give the spooks an effective tool for revealing predicting IED activity--and identifying the networks that support it--allowing war fighters to target the "system" before the bombs go off.


POW said...

Sorry I don't use Blogs. Just wondering how you got a hold of the RADII whitepaper.

BrianFH said...

Ideally, this sort of tool generates a "virtuous circle", wherein the use of the system by trained analysts and on-site users improve the tool and its data sets, which leads to more and better use and results, which improves the algorithms, etc.

It's a multiplier, and permits things to be done in a short time that would have been impossibly onerous before.