Motion-based Routing for Opportunistic Ad-hoc Networks

摘要

In this paper, we introduce novel motion-based routing protocols for opportunistic packet relay in ad-hoc opportunistic networks. Unlike existing ad-hoc routing protocols, which require global knowledge about the mobility patterns of other nodes, thus raising scalability and privacy concerns, our routing protocols store and use only local motion information, such as the current direction of the local node. Such information is easily obtainable by any modern GPS-equipped mobile device. We design and evaluate a series of protocols using different motion information, ranging from simple, easy to capture metrics i.e., speed and direction, to more complex metrics, i.e., the past or expected trajectory of the local node. These metrics allow for increasing degrees of routing accuracy at a correspondingly higher cost in terms of memory and computation. However, all of our routing protocols have constant complexity in the size of the network, and none of them rely on nodes sending or storing information about other forwarding nodes. This makes our schemes ideal for large-scale networks and non-community networks, where membership is very dynamic. Our simulation study using real-world traces shows that our motion-based schemes perform comparably to or considerably better than existing complex, global-knowledge-based protocols with significant less utilization on memory, CPU, and bandwidth.