LPS and LRF: Efficient buffer management policies for Delay and Disruption Tolerant Networks

Abstract

In Delay and Disruption Tolerant Networks (DTNs), the message delivery rate is impacted by the buffer management policy adopted by nodes once buffer overflows occur frequently. This paper proposes two new buffer management policies. The first one, called LPS (Less Probable Sprayed), uses the messages delivery probability and estimates the number of replicas already disseminated to decide which message to drop. The second one, named LRF (Least Recently Forwarded), drops the least recently forwarded message based on the assumption that messages not forwarded over a certain period of time have already reached several next hops. These two policies are implemented and compared with existing proposals found in the literature. The analysis considers traces of three real networks in which LPS and LRF policies provide higher delivery rates up to 75% and 37%, respectively, than the rate provided by the second best policy, with less overhead.