Fast fault detection in wireless sensor networks

Fault detection is of a high importance, because if the fault diagnosis is not done, it is possible that gained data would not be correct and so would be decision making. Perfect and safe sensors and defected sensors, should be separated from each other. In this paper we have presented a distributed algorithm to detect the faulty sensors, This method is based on neighbor voting and it acts as each sensor, according to neighbor voting, makes decision itself. In this method, each sensor sends its state with the value sensed from the environment to neighbors alternatively. All sensors based on information received from neighbors about whether the sensor is impaired or not, make decisions. In this algorithm, three parameters of θ1 and θ2 and θ3, which determining their accurate value is very important, by any change in those, the result would change. Increasing the amount of θ2, the faulty sensor detection accuracy (FSDA) reduces and the false alarm rate (FAR) increases. One of the differences of This approach to previous method [1], is that the sending position would be with data, That itself makes the algorithm run faster and also it reduces the amount of sending information that as the result, the energy consumption.