Rapid Deployment with Confidence: Calibration and Fault Detection in Environmental Sensor Networks

Rapidly deployable sensor networks are portable, reusable, and can take advantage of a human user in the field attending to the deployment. Unfortunately, even small disruptions or problems in collected data must be addressed quickly, as the overall quantity of data gathered is small relative to longterm deployments. In this paper we describe a procedure for calibration and a system for online fault remediation. Care in the calibration process for ion selective electrodes used for water quality assists interpretation of the data. Scientists will have more confidence in the data obtained from a rapid deployment if in-field users can detect and compensate for problems as they occur. We have designed and implemented a tool for use in the field to detect potential faults and provide actions to remedy or validate the faulty data. In January of 2006 we deployed 48 sensors over a period of 12 days in Bangladesh in order to aid in validating a hypothesis on the mass presence of arsenic in the groundwater. Our system is based on the the approximately 25,000 measurements we collected.