Hierarchical Routing in Sensor Networks Using k-Dominating Sets

For a connected graph, representing a sensor network, distributed algorithms for the Set Covering Problem can be employed to construct reasonably small subsets of the nodes, called k-SPR sets. Such a set can serve as a virtual backbone to facilitate shortest path routing, as introduced in [40], [12] and [13]. When employed in a hierarchical fashion, together with a hybrid (partly proactive, partly reactive) strategy, the k-SPR set methods become highly scalable, resulting in guaranteed shortest path routing with comparatively little overhead. In this paper, we first discuss the notion of k-SPR sets, with the nodes of such a set functioning as routers for the network. These sets generalize our earlier k-SPR sets, which facilitated shortest path routing. We then introduce K-SPR sequences that are used for hierarchical routing. We propose a distributed greedy algorithm for construction of K-SPR sequences. The new sets facilitate minimal path routing, where “minimal path” here means “shortest weighted path based on edge weights”. Finally, we introduce an efficient hybrid hierarchical routing strategy that is based on K-SPR sequences. Our approach is unique in the sense that although dominating sets have been used to construct virtual backbones in ad hoc and sensor networks, this is the first attempt to use k-hop connected k-dominating sets for hierarchical routing that is also minimal path routing.