A distributed clustering algorithm for fault-tolerant event region detection in wireless sensor networks

Wireless sensor networks which are envisioned to consist of many simple processing, storage, sensing, and communication capabilities are believed to open the doors to a plethora of new applications. Efficient robust data aggregation is a key feature in information processing in wireless sensor environments, especially in the presence of faulty sensor nodes which arise due to harsh environments or manufacturing reasons. These conditions highly affect the quality of gathering data in different locations and times. In this paper we propose a distributed localized method for detecting regions that are susceptible to message loss above a given threshold, and then based on this method we suggest an algorithm for clustering nodes in that region which proportionally adapts based on message loss. Simulation experiments confirm the validity of the proposed algorithm with a high degree of accuracy.