Ant Colonies Offer Network Solutions

Anyone who has spent time gazing at a colony of zigzagging ants has probably noticed the insects are all about teamwork.

Working together, a swarm of ants can haul a piece of food 10 times their size up a steep slope. And, somehow, workers streaming to and from the nest always seem to settle on the shortest path to a food source.

Recently engineers have taken notice of the insects’ impressive use of collaboration and have started finding ways to apply it to problems in the world of humans.

“Ants have been around for 50-90 million years,” says Eric Bonabeau, a telecommunications engineer and biologist who conducted studies on ant colonies at the Sante Fe Institute. “That might be the reason why they’ve got a good system down that doesn’t require complex units.”

Collective Intellect

The ant, itself, Bonabeau points out, is not a complex unit. In fact, all of its movements are based on immediate reactions to its surroundings or to its fellow ants. Put those ants together, however, and a sophisticated system emerges. Some scientists call it a collective intelligence.

Take, for example, the ability of ants to find the shortest path to a food source. When an obstacle, such as a stick or even a person’s foot, blocks the most direct path, ants very quickly find the next best route.

Translate that ability to glitches on the Internet or roadways or in telephone lines and the ant can offer some solutions. If the nodes on one Internet network are clogged with too much traffic, it’s sometimes necessary to reroute new traffic. The same problem occurs with telephone lines that become tied up or trucking routes that become congested by holiday traffic.

Ants get around the problem by laying down a thin layer of signaling chemicals called pheromones wherever they travel. When other ants detect these pheromones, they instinctively follow the path the chemicals mark. The thicker the pheromone trail, the more likely other ants will follow the path.

Because the ants that follow the shortest path are also those first to make a return trip to the food source, their pheromone trail quickly becomes thicker. The heavier pheromone scents attract more ants and the shortest path is even further reinforced.

Meanwhile, there are always some ants that follow their own trail and explore new routes. These individuals also lay pheromone trails as they go. So when, say, a rock tumbles across the main route and traffic is jammed, the ants are ready with a backup path. Rarely used, inefficient routes are eventually abandoned as the pheromone trails marking them evaporate.

“Ants are not adaptive, themselves,” explains Marco Dorigo, a computer scientist at the Universite Libre de Bruxelles in Brussels, Belgium, and coauthor with Bonabeau of a study on ant swarming in this week’s issue of Nature. “It is the ant colony as a whole that adapts to the changing problem.”

To build on the technique that they observed in ants, Dorigo and Bonabeau devised a system they call the Traveling Salesman Problem. In this scenario, virtual ants travel to every point on a given electronic map. When they reach a node that is highly useful (by offering say, a traffic-free zone or a powerful connector) they are programmed to release more virtual pheromones. Other virtual ants then follow this preferred route and eventually the best path is mapped out on the network.

Bonabeau says the system is already being used to design clog-free telephone and computer networks and to map efficient trucking routes in Switzerland.

Heavy Lifting

Another ant trick that has inspired Dorigo and Bonabeau is the way the insects team up to carry improbably large objects up steep slopes. Human movers carrying, say, a couch up a stairway might handle such a problem by talking each other through the task. But ants communicate through the object they’re carrying.

“They don’t exchange information directly,” says Bonabeau. “Instead, if the item needs realignment, ants will sense that and reinforce the weak side.”

The same idea has now been applied to robots that are designed to move large boxes. The robots are able to manipulate boxes and move them efficiently without actually communicating with one another.

Although Bonabeau is a biologist as well as an engineer, he stresses that his main interest lies in finding solutions in nature while not necessarily understanding exactly how nature works. One of his ant swarm models, for example, is based on a system that he isn’t sure actually exists in ant colonies. Instead, he drew from what he knows about ant swarms and devised a system that could theoretically work in an ant colony.

“We don’t necessarily care about biological accuracy,” he says. “What we care about is taking what we know about species like ants and creating efficient systems.”

Eamon Mallon, a researcher at the Ant Lab at the University of Bath in England says that Bonabeau’s approach is understandable, considering how difficult it is to observe ant swarms carefully.

Paint by Ants

To study even very small colonies of about 200 ants, Mallon and his colleagues first individually mark each ant. To do this painfully minute work, they anesthetize the ants using carbon dioxide. Then they use single hair paintbrush's and model paint to make four distinctive markings on each three-millimeter ant.

“You make sure your hands aren’t shaking and you haven’t drunk too much coffee and then you paint your ants,” Mallon says, adding that “people can’t believe I do this for a living.”

Once each ant is marked, Mallon and his colleagues can trace what each ant is doing. Right now, Mallon is learning how ant groups settle on the best locations for nests.

While Bonabeau and Dorigo have focused on applying ant colony behavior to automated systems where each unit is simple and predictable, Mallon believes that ant systems can even work for settings that involve one of the most complicated agents — humans. He says that ant colonies offer good models for designing buildings, office work-flow arrangements and crowd control.

“Some are quick to point out that people are much more complicated than ants,” he says. “But in a group, people aren’t thinking many existential thoughts. They’re more like ants. They’re just trying to figure out where to go.”