Lightweight Fault Detection Strategy for Wireless Sensor Networks Based on Trend Correlation

In this work, we propose a fault detection strategy for wireless sensor networks (WSNs) called Trend Correlation based Fault Detection (TCFD). This can detect the faulty nodes through analyzing trend correlation and median value of neighboring nodes. On basis, aiming to avoid excessive routing overhead caused by over-frequent detection, self-starting mechanism is designed on cubic exponential smoothing method. Since results in TCFD are determined historical data at consecutive times do not rely comparison instantaneous sensed values single moment, it significantly reduce impact time accuracy. The simulation have indicated that compared with referenced strategies, proposed obtain better accuracy four common types nodes; case where real rate network reaches 0.5, least 70% be detected false alarm still kept below 30%; help mechanism, response faults shortened.