2005 BES Annual Meeting Presentation and Poster Abstracts
A Wireless Sensor Network for Soil Monitoring
Josh Cogan, Katalin Szlavecz, Razvan Musaloiu-E., Andreas Terzis, Alex Szalay, Sam Small
Abstract: The most spatially complex stratum of a terrestrial ecosystem is its soil. Among the major challenges of studying the soil ecosystem are the diversity and the cryptic nature of biota, and the enormous heterogeneity of the soil substrate. Often this patchiness drives spatial distribution of soil organisms, yet our knowledge on the spatio-temporal patterns of soil conditions is limited. To monitor abiotic conditions in the soil at biologically meaningful spatial scales we have been developing a wireless sensor network. Each node is based on a MICAz mote connected to a custom-built sensor suite that includes a Watermark soil moisture sensor, an Irrometer soil temperature sensor, and sensors capable of recording ambient temperature and light intensity. We developed the software running on the motes from scratch, using the TinyOS development environment. Each mote collects measurements every minute, and stores them persistently in a non- volatile memory. The decision to store data locally at each node enables us to reliably retrieve the data in the face of network losses and premature node failures due to power depletion. Collected measurements are retrieved over the wireless network through a PC-class computer acting as a gateway between the sensor network and the Internet. Our software powers down sensors between sampling cycles, and turns off the radio (the most energy prohibitive mote component), when not in use. By doing so we were able to increase node lifetime by a factor of ten. We collected field data over several weeks. The data was ingested into a SQL Server database, which provides data access through a .NET web services interface. The database provides functions for spatial and a temporal interpolation of the data.