Imagine a colony of ants. One of the ants is walking around with a notepad. This ant needs to send a message to a fellow ant, so it writes the message and hands it over to the next ant that it encounters. All of the ants move in a random manner, so no ant can tell up front where the recipient ant will be, or when they will meet. Since this second ant is as likely to encounter the intended recipient as any other ant, it will hold onto the message until it does. To simplify the analogy, each ant can only carry one note.

The ant’s message has only a limited lifespan; so although every pair of ants will eventually meet at some point in the future, the authors are interested in the probability that the message will arrive at its destination within a set time period, which I will call the “success probability.” What if the message is too big to fit on one piece of paper, and needs to span multiple pieces? Obviously, this reduces the probability that the recipient will receive the full message within the required time limit.

The authors evaluate the success probability when they factor in replication using erasure codes (the source ant hands out duplicate copies of the notes) and fountain codes (each note is some random combination of the whole message) within an energy constraint (the source ant has a limited amount of paper to write on). The authors do not use ants to describe delay-tolerant networks, but instead use calculus, statistics, and simulation; however, the analogy works well for this brief description.