Towards effective fault detection in heterogeneous wireless sensor networks

Abstract

Accurate detection of the faulty nodes and transmitting correct status data of a monitored environment is one of the crucial challenges in wireless sensor networks (WSNs). This paper puts forward a novel unsupervised faulty node detection algorithm to improve the detection accuracy of events in the 2D and 3D heterogeneous WSNs. In this approach, the current status of each node is predicted by using the ARIMA methods and factors such as distance and the common coverage ratio. Moreover, the node's status in the previous rounds is employed to weight the measured values of the neighbouring nodes. By using the proposed distributed algorithm, each sensor node can recognise its status with more accuracy, in the presence of transient faults and events like fire. Extensive simulations indicate the effectiveness of this algorithm in reducing the false positive problem and improving the detection accuracy in different scenarios considered for the 2D and 3D heterogeneous WSNs.