Fault detection in wireless sensor networks

Abstract

Wireless Sensor Network (WSN) deployment experiences show that data collected is prone to be imprecise and faulty due to internal and external influences, such as battery drain, environmental interference, sensor aging. An early detection of such faults is necessary for the effective operation of the sensor network. In this preliminary work, we propose a hybrid approach to the detection of faults and we illustrate its performance on data coming from a real sensor deployment. The proposal is a first step to have a hybrid method towards automated on-line fault detection and classification in context-aware WSNs middleware framework.