Two Effective Attack Detection Algorithms in Wireless Sensor Networks

Based on the model of resilient data aggregation in wireless sensor networks, two attack detection algorithms by means of % -testing were proposed in this paper according to clustering uniformly or not, namely the CTUC (chi-square test for uniform cluster) algorithm and CTUUC (chi-square test for un-uniform cluster) algorithm. By applying % -testing with some priori knowledge about the system, the novel algorithms may effectively detect the attacking behaviors in the network and have a higher detection probability compared with the SC (split and check) method presented in the literature. In addition, the CTUC algorithm is suited well to the situation of an unknown distribution function or a larger variance of the sample data, while the CTUUC algorithm has a better detection performance when the variance is smaller. Simulation results show the effectiveness of the presented algorithms.