Anternet: The regulation of harvester ant foraging and Internet congestion control

Distributed networks, such as ant colonies and the Internet, operate without central control. Such systems are regulated using feedback based on local interactions among simple components. In this paper, we investigate the algorithms evolved by harvester ant colonies in Arizona to adapt the foraging rate to the available food on a given day. The connection with congestion control algorithms in the Internet, namely TCP, is studied. We find that the algorithm used by the ants corresponds with the three main states of TCP - slow-start, congestion avoidance and timeout. We obtain a mathematical model of the feedback based steady-state foraging activity of the ants. We also consider the spacings between successive returning foragers and find that this process is well-approximated by a Poisson process at high foraging rates.