Energy and delay optimal epidemic relaying in delay tolerant networks

In this paper we introduce a novel framework for the distributed control of DTNs in order to support message replication the devices acting as relays need to sacrifice part of their batteries. Due to the fact that feedback messages in DTNs cannot be used, the knowledge of the delivery status is unknown for the source. We model the problem as a Partially Observable Markov Decision Process (POMDP) and we derive an optimal policy for forwarding messages in the network; in fact the source node uses a belief to decide either to continue forwarding a message or not. We provide extensive numerical results to validate the proposed scheme in which our mechanism exchanges less messages than epidemic routing, hence, less energy is consumed.