Effect of neighborhood on in-network processing in sensor networks

Abstract

Wireless sensor networks are growing from a few hand-placed devices to more large-scale networks in terms of coverage and node density. For various concerns, such as scalability, larger network sizes require some management of the large volume of data that a sensor network delivers. One way to manage this data is processing information in the network. This paper investigates how a sensor network's network architecture (specifically, the neighborhood structure) can influence the conclusions that a sensor network makes from its measurements. The results demonstrate that non-planar structures are infeasible for routing and some in-network processing applications. Structures with low average edge lengths give better quantitative results, while those with high edge densities give better qualitative results. © 2008 Springer-Verlag Berlin Heidelberg.