All delay-tolerant networks (DTNs) suffer from a shortcoming: they are unable to sustain continuous connectivity between neighboring nodes. The problem of successful message forwarding in such networks is analogous to obtaining multiple solutions for a multicriteria optimization problem.

The authors of this paper propose a protocol--cultural greedy ant (CGrAnt)--that ascertains successful message delivery by providing dynamic labels to intermediary nodes of a DTN, based on their past success in forwarding messages to the intended destinations. The proposed protocol is a combination of a greedy variant of the ant colony optimization metaheuristic and a cultural algorithm. The underlying concept introduced in the paper is the use of data messages to encapsulate forward ants in order to assess individual routes within the DTN and provide appropriate grading to intermediary nodes; this information helps determine whether they may (or may not) be selected for subsequent message forwarding on a particular route.

The main feature of the proposed protocol is information storage, or best route path detail, that is not centralized on a single node, but rather is distributed across the network. Another feature is that the belief space associated with routing paths possesses three types of knowledge: situational, domain based, and historical.

In general, the proposed scheme is a very interesting contribution that will help improve message delivery in DTNs, as is evident from the analysis of simulation findings and the comparison of it with the performance of other schemes, as reported in the paper.