Neighborhood-Based Topology Recognition in Sensor Networks

TL;DR

This paper presents algorithms for sensor networks to recognize their environment's topology and boundaries using local communication, without location hardware, relying on stochastic, topological, and geometric methods.

Contribution

It introduces novel algorithmic approaches for boundary detection and topology recognition in sensor networks with limited communication capabilities.

Findings

Effective boundary detection with limited local communication

Methods for computing Voronoi diagrams and network thickness

Algorithms applicable to densely distributed sensor nodes

Abstract

We consider a crucial aspect of self-organization of a sensor network consisting of a large set of simple sensor nodes with no location hardware and only very limited communication range. After having been distributed randomly in a given two-dimensional region, the nodes are required to develop a sense for the environment, based on a limited amount of local communication. We describe algorithmic approaches for determining the structure of boundary nodes of the region, and the topology of the region. We also develop methods for determining the outside boundary, the distance to the closest boundary for each point, the Voronoi diagram of the different boundaries, and the geometric thickness of the network. Our methods rely on a number of natural assumptions that are present in densely distributed sets of nodes, and make use of a combination of stochastics, topology, and geometry. Evaluation requires only a limited number of simple local computations.