A Markov chain model for drop ratio on one-packet buffers DTNs

Abstract

Most of the efforts to characterize DTN routing are focused on the trade-off between delivery ratio and delay. Buffer occupancy is usually not considered a problem and most of the related work assumes infinite buffers. In the present work, we focus on the drop ratio for message forwarding considering finite buffers. We model message drops with a continuous time Markov chain (CTMC). To the best of our knowledge, there is no previous work with such approach. We focus on the worst case with 1-packet buffers for message forwarding in homogeneous inter-contact times (ICT) and 2-class heterogeneous ICT. Our main contribution is to link the encounter rate(s) with the drop ratio. We show that the modeled drop ratio fits simulation results obtained with synthetic traces for both cases.

Using emulation to validate applications on opportunistic networks

“…The main goal of analytical models is to provide a closed formula for a specific characteristics. Others propose algorithms to approach reality [17,18,19,20]. However, most of these models assume unrealistic hypotheses or they cannot scale.…”