Saturday, March 29, 2014

The Anternet

Ant colonies operate coherently without central control. They can offer insights on other such systems that operate without central control, such as neurons in our brain, or cancers. Deborah Gordon gave a fascinating talk exploring these relationships.

There are over 12,000 species of ants and they operate under different algorithms that strike a balance between operating costs and resource availability, depending on their environment.

For instance, in the desert, operating costs are high, since ants have to expend precious water to go out foraging for food and more water. In this environment, ants don’t leave to forage until there’s been a positive signal, namely a certain frequency of ants returning from foraging. Since ants only return once they’ve reached their foraging target, a steady stream of returning ants indicates foraging has been successful. Different colonies have different trigger. Because of advances in genetic decoding which allows identification of offspring colonies, Gordon has been able to analyze many years worth of data to show that higher trigger levels yield better evolutionary success.

The Internet works in a similar way to these desert ants. Data doesn’t leave an Internet node until there is a positive signal that there’s enough bandwidth available. Hence the name the Anternet. The expression “to a hammer, everything looks like a nail” comes to mind at this point.

Ants in the tropics have an opposite algorithm to desert ants. The environment is friendly, so operating costs are low, but resources can be scarce because of the plethora of competition. So these ants go out foraging all the time, until they receive a negative signal. Understanding this behaviour in ants could yield new insights for co-operating robots.

When ants find a really juicy stash of food, they recruit other ants to join the party, so to speak, thus explaining that steady stream of ants attacking your last picnic. It’s possible that cancer cells have a similar recruitment behaviour and so understanding ants might even provide some insights into cancer.

What an engrossing field. Ants may not be as pretty as fireflies (Sara Lewis' focus as described here), but, to me, they're immensely more fascinating.

No comments: