MIT: How do the ants decide where they will move their nests? Can this particular question teach one of the largest technology universities? Computer scientists are looking for effective ways to collect data from distributed sensor networks, and according to MIT researchers ants can help.
It turns out that the frequency with which ants explode and collide with each other as they roam about looking for a new colony is a very good indication of how researchers should look for.
When many ants gather in a nest, the workers in the ant colony pick up the queen and carry her to the chosen location, says Cameron Musco, an MIT graduate student in the electrical and computer science department.
The researcher with team , he examined the mathematics behind ant behavior in search of an efficient way to estimate population density.
The peritreatment and potential collisions in what researchers call random walks, gives ants a sense of how interested a critical mass of other ants are in finding a new colony.
Musco and his colleagues simulated the behavior of ants “with a mathematical model of random walks in which a region is divided into a grid of nodes, and each node is connected to a series of other nodes. In the random walk the model of ants they created could move to any other node connected to the one they are on, including the one they left.
Relatively quickly, the frequency with which ants run provides a reliable estimate of how many other ants are in the area, according to the algorithm developed by the team.
Researchers report that random walks can produce estimates about as accurate as random sampling, by which they poll a series of randomly selected nodes in a network. This makes the method very useful in networks where it is impossible to try randomly because there is no complete list of network components.
For example, it could be used if one wanted to find the political beliefs of members of a social network for which no list of members is available, making it impossible to create a random sample. A random walk could give the information by going from one person to that person's contacts, their contacts, and so on, counting, say, Democrats and Republicans.
Read more:
http://news.mit.edu/2016/ant-colony-behavior-better-algorithms-network-communication-0713
