DTN routing strategies using optimal search patterns

Biologists have long shown that the mobility patterns of many foraging animals and insects are similar to Levy walks and Levy walks are an optimal search strategy when target objects (i.e., food sources) are sparse and their locations are not known in advance. In this paper, we apply Levy walk patterns to routing in delay tolerant networks (DTN). In DTNs, message forwarding nodes often do not have full information about the whereabout of message destinations. Using the optimality property of Levy walks, we devise two styles of routing strategies. One is an active strategy using message ferries (MF) where the movement of MFs can be controlled to have a Levy walk pattern in order for them to maximize the opportunity of meeting the destinations and the other is a passive strategy in which the movement of nodes cannot be controlled, but messages are forwarded in such a manner that their forwarding patterns mimic the Levy walk patterns. We show through simulation that (1) both strategies are very effective when knowledge about destinations (i.e., contact history, trajectory or locations of destinations) is highly limited and (2) they complement existing utility-based routing which excels when such knowledge is available.