Algorithmic Aspects of Large Sensor Networks

A basic scenario for organizing a large number of nodes into a functioning sensor network can be described as follows. Given a large swarm of immobile sensor nodes that have been scattered in a polygonal region, such as a street network; nodes have no knowledge of size or shape of the environment or the position of other nodes; moreover, they have no way of measuring coordinates, geometric distances to other nodes, or their direction. Their only way of interacting with other nodes is to send or to receive messages from any node that is within communication range. The objective is to develop algorithms and protocols that allow self-organization of the swarm into large-scale structures that reflect the structure of the street network, setting the stage for global routing, tracking and guiding algorithms. We present an approach that works in two stages: boundary recognition and topology extraction. All steps are strictly deterministic, yield fast distributed algorithms, and make no assumption on the distribution of nodes in the environment, other than sufficient density. In order to develop and test useful algorithmic concepts, we have created our own network simulator: Shawn, which aims at providing an infrastructure for testing high-level algorithmic ideas. While traditional simulators like tt ns-2 reach their limits at about 2,000 nodes, Shawn can easily simulate networks with several hundreds of thousands of nodes. In addition, its geometry-oriented data structures aim at specifically supporting updates and activities that are present in networks with mobile nodes. Using graphical simulation output created by Shawn, we demonstrate the power of our concepts by showing how a network with 60,000 nodes can self-organize into a set of clusters that reflect the topological structure of the node swarm.