Impact of Super-Diffusive Behavior on Routing Performance in Delay Tolerant Networks

Abstract

Motivated by the recent findings of super-diffusive patterns in mobility traces, we investigate the impact of super- diffusive behavior of mobile nodes on contact-based metrics and performance metrics of routing protocols in delay tolerant networks (DTNs). We show that diffusive properties make huge impact on the performance of routing protocol - message delivery ratio and delay of delivered messages, and existing models such as random waypoint models or Brownian motion models lead to overly optimistic or pessimistic results when diffusive properties are not properly captured. In addition, we point out that existing contact-based metrics are unable to differentiate between varying degrees of routing performance under different diffusive mobility patterns, and then propose to use the number of new contacts as a far more effective metric, especially for scenarios in which message routing/forwarding is built upon contacts among mobile nodes. Our work in this paper suggests that the diffusive behavior of mobile nodes should be taken into account, for the design and the performance evaluation of network protocols in DTNs.