Abstract:

A fault detection algorithm based on Dirichlet reputation for wireless sensor networks was proposed. The algorithm consists of three modules: measurement, reputation updating and fault judgment. The healthy level of sensor node in current time was evaluated in the measurement module. The work of maintaining and updating the reputation of the sensor node were performing by the reputation updating module. Fault judgment module determines whether a node is faulty or not, and whether the sensitive event happens or not. Simulation results show that the proposed algorithm has a high fault detection rate. Even 25 percent of nodes are faulty, 94.83 percent of fault nodes can be detected out and the false alarm rate is below 5%.

Key words: wireless sensor network, fault detection, reputation, Dirichlet distribution