Common-Cause Sensor Fault Detection and Robust Data Fusion in Distributed Wireless Sensor Networks

Reliable sensor measurements and accurate data fusion are essential for distributed wireless sensor networks (WSNs). Network reliability, however, can be greatly compromised by independent sensor failures, common cause sensor failures, and correlated measurement errors. In this study, a Common-Cause-Aware Fault Detection and Robust Data Fusion (CCF-RDF) scheme is proposed that combines common-cause fault identification, adaptive assessment of sensor reliability and robust data fusion with spatiotemporal correlation analysis. Evaluation using simulation takes into account different sensor fault conditions, network size, measurement noise, and packet loss rates. The illustrative results show that the proposed framework can attain a Fault Detection Accuracy (FDA) of 97.35%, False Alarm Rate (FAR) of 3.21%, and Maximum Common-Cause Fault Identification Rate (CCFIR) of 95.18%. The FRMSE values are 0.0385, 0.0654 and 0.0928 for normal, independent fault and common cause fault condition, respectively. The Network Energy Consumption (NEC) rises from 16.32 J to 47.25 J as the number of nodes in the network grows from 50 to 150 nodes. The results indicate the possibility of incorporating correlation information to reassess the reliability in distributed WSNs for fault detection and robust data fusion.