Fault Data Diagnosis by Cluster Computing in Wireless Sensor and Actuator Networks

Since sensors and communication links are prone to fail, to propose an efficient fault diagnosis algorithm becomes an important issue in wireless sensor and actuators networks. However, most of researches focused on the link fault tolerance without considering the sensing fault tolerance. Actuators may perform incorrect actions on receiving the fault data because the sensing function is on malfunction but the communication function is capable. Therefore, a fault data diagnosis by cluster computing (FDDCC) was proposed in this paper. In FDDCC, each sensor was clustered according to an actuator. Each actuator acted as a cluster head by a clustering mechanism. Each actuator selected the correct data and detected the fault data by cluster computing in a distributed manner. Simulation results showed that FDDCC had the better performance than other fault diagnosis algorithms, such as distributed fault detection method (DFDM), even if the ratio of the fault data to all data or the sensor density varied.