Two-Tier Framework for Sensor Fault Characterization in Sensor Networks

For a sensor network to be reliable and useful, sensor data must sustain high Quality of Information (QoI). While QoI depends on many factors, the most crucial is the integrity of sensor data sources themselves. Sensor data quality may be compromised by causes such as noise, drifts, calibration, and faults. On-line detection and isolation of such misbehavior are crucial to assure high QoI for the end-user, and efficient management of network resources. We describe a two-tiered system for on-line detection of sensor faults. A local tier running at resource-constrained nodes uses an embedded model of the physical world with a hypothesis-testing detector to identify potential faults and notifies a global tier. In turn, the global tier uses these notifications for consistency checking among sensors and provides more robust estimates for events of interest, and also generates feedback to update the local models. We demonstrate the performance of our system by investigating its impact on the application QoI.