An efficient prediction-based routing protocol in delay tolerant networks

Delay Tolerant Networks (DTNs) are characterized by sparse node density, uncertain node mobility and lack of global information, which make routing one of the most challenging problems. In this paper, we propose a new routing protocol for DTNs. We observe that the forwarding performance of a node is not only determined by its contact schedules with the destination but also affected by its contacts with the neighbor where the packet is received from, which has not been considered in most of existing routing schemes in DTNs. Based on this observation, we design a novel routing metric, called Instant Delivery Probability (IDP), which provides an accurate estimation on node forwarding performance in terms of packet delivery ratio and can be efficiently calculated with local information. The single-copy and multi-copy forwarding algorithms are also presented, where each message is opportunistically forwarded to the nodes with largest IDP to maximize the delivery probability. Extensive trace-driven simulations show that our routing protocol with IDP significantly improves the routing performance compared to the state-of-the-art forwarding strategies.