Data clustering-based fault detection in WSNs

Sensors easily become faulty and unreliable subject to limited battery and insecurity. Data Fault is one of traditional faults in the wireless sensor networks. Data fault mainly uses distributed method through exchanging neighbors´ measurements and voting for decision. But the detection accuracy performance is easily influenced by unbalanced fault distribution. Based on this, we propose the k-means clustering-based fault detection algorithm (k-CFD), which uses clustering view to replace tendency values for fault decision, in addition, and adopts ant colony optimization algorithm to promote the results of k-means mechanism. The simulation results demonstrate the efficiency and superiority of k-CFD mechanisms.