Geosensor Networks

Advances in miniature, low-cost microelectronic and mechanical systems (MEMS) with limited onboard processing capabilities, storage, and short-range wireless communication links, together with the development of novel microsensors and sensor materials, enable us to build a new generation of technology that consists of large collections of tiny, untethered, battery-powered computing nodes with various sensing functions. With the continued trend toward miniaturization and the inexpensiveness of such sensor nodes, it is expected that sensor nodes will be [Page 1253]less than a cubic millimeter in size in the near future and that sensor networks can be made up of thousands or even millions of sensors (“smart dust”). Geosensor networks (GSNs) are a specialized application of wireless sensor network technology in geographic space to detect, monitor, and track environmental phenomena and processes at a novel spatial and temporal scale. Considering remote sensing instruments such as “telescopes,” which are commonly used to monitor environmental processes on Earth, a GSN can be viewed as an “environmental microscope” providing a spatiotemporal resolution of observations and near-real-time information never available before.

The technology of GSNs has the following constraints that pose new challenges from an infrastructure system and application development standpoint:

Since the mid 1990s, much research in sensor networks has focused on the design of tiny computing platforms, operating systems, and programming languages for resource-constrained computing environments (e.g., TinyOS, Contiki, and nesC). A major task of GSNs is intelligent data collection and processing. Despite their powerful and novel capabilities to observe the physical world, programming sensor networks for specific observation and actuation tasks is cumbersome today due to the failure-prone nature of nodes and communication links and the vast parallel computational nature of such systems. A domain scientist interested in deploying a GSN needs to be able to define the necessary tasks in a user-friendly way and delegate the optimization and ultimately self-adaptive execution to the run-time system without having to worry about the details.

...