Combining on-demand and opportunistic routing for intermittently connected networks

While current on-demand routing protocols are optimized to take into account unique features of mobile ad-hoc networks (MANETs) such as frequent topology changes and limited battery life, they often do not consider the possibility of intermittent connectivity that may lead to arbitrarily long-lived partitions. In this work, we introduce the space-content-adaptive-time routing (SCaTR) framework, which enables data delivery in the face of both temporary and long-lived MANET connectivity disruptions. SCaTR takes advantage of past connectivity information to effectively route traffic towards destinations when no direct route from the source exists. We show through simulations that, when compared to traditional on-demand protocols, as well as opportunistic routing (e.g., epidemic), SCaTR increases delivery ratio with lower signaling overhead in a variety of intermittently connected network scenarios. We also show that SCaTR performs as well as on-demand routing in well-connected networks and in scenarios with no mobility predictability (e.g., random mobility). In the latter case, SCaTR delivers comparable reliability to epidemic routing with considerably lower overhead.