Fragmented Data Routing Based on Exponentially Distributed Contacts in Delay Tolerant Networks

Abstract

Existing works in data forwarding in delay tolerant networks often simplify the data transmission during a contact between two network nodes. One common assumption is that once transmitted, messages will always successfully arrive at the encounter node in its entirety regardless of the contact duration. The other common assumption is that messages are unfragmented, and are successfully transmitted only if the contact duration is long enough. If the contact duration is too short, the transmission will not complete, and the message needs to be re-transmitted in its entirety in the next contact. While the first assumption is unrealistic, the second assumption makes the routing strategy inefficient for large data items in network environments where most contact durations are short. In this work, we eliminate the assumption of unfragmented data, and propose a single-copy fragmented data routing strategy (FDR) that performs well with short contact durations. FDR considers the exponential distribution of inter-contact time, contact duration time, and fragmented data sizes when making forwarding decisions. We show that under short contact durations and large message sizes, our scheme can achieve up to 37% higher delivery rate and 35% lower delay compared to other routing strategies.