Security
6:00 pm
Tue August 2, 2011

Ants might become the new model for cybersecurity

RICHLAND, Wash. – Ants have an incredible ability to track down food and swarm together against enemies. Now, scientists at the Pacific Northwest National Lab are hoping to use the same model to fight off cyberattacks.

The new tools are called digital ants.

Around the office, Glenn Fink is known as the "Ant Daddy."

"My specialty is in cyber security and visualization," Fink says. "I'm the guy who came up with the idea of applying the designs of ants to cybersecurity."

Digital attacks have been a sore subject at the Pacific Northwest National Lab lately. In early July, someone launched a cyberattack against the federal research center in southeast Washington. The huge lab known for its expertise in cybersecurity suddenly had to shut down its email, Internet and other research systems for a week.

Programs act like ants

Long before that incident, Glenn Fink was shopping around for a new nature-based model to fight cyberattacks. Bees, termites, slime mold ...  nothing quite fit until ants.

"Ants can find the shortest path to food even when the path to that food changes. That sort of resilience was what we were looking for," Fink explains. "So when we saw that we said, 'It's got to
be ants.'"

Here's how they work: Digital ants are little programs that Fink sets loose in a computer network. Their job is to wander around and look for things that look suspicious, like too many hits at night on one site or a program that doesn't shut down when its supposed to.

That ant marks the problem. That attracts other ants. If they also think it's a problem they mark it too, which attracts yet more ants. Eventually, the problem looks like a digital ant hill, prompting a human to check it out, all in a matter of seconds.

Hunting symptoms

Fink says the beauty of this program is that the ants are not looking for a specific virus or problem. Rather, the ants are looking for symptoms of intruders, which are much harder to hide.

"We're looking for outside indicators that just happen to indicate something is wrong," he says. "It's kind of hard for people who are designing malicious programs to avoid
these side effects no matter what they are."

But both Fink and his colleague David McKinnon say ants aren't fail proof. Someone could attack the ants themselves or take over the ants' brains.

Applications

McKinnon studies ways to get the ants to protect huge online systems like the nation's power grid. He says, just like real ants at a picnic, digital ones are hard to get rid of.

"The colony is going to win. They are going to come and take away the birthday cake or whatever they have at the picnic," McKinnon says. "Yeah, a few of the ants might be squished on the table cloth but collectively they are going to manage to do their task."

So far, Fink says digital ants are still in the study phase. "We need to make digital ants and other systems more like natural systems so they can work with humans, who are a natural system themselves," he says.

Copying nature like that is complicated. He thinks it will be at least five to 10 years before he can deploy his digital ants.

On the Web:

Copyright 2011 Northwest Public Radio