Research on Intrusion Detection in WSN Based on Projection Pursuit

This paper proposes a small calculation and high-efficiency property for detecting malicious attacks in wireless sensor networks. We use projection pursuit which turns high-dimensional node properties to low-dimension space and all node's attributes not only accurately but also clustered easily. Our method effectively solve the problem of low efficiency of the large amount of computation of high dimensional data. We use ten nodes include normal nodes and abnormal nodes to test our model. We prove that through the projection pursuit, we calculation of the value of each node and observance of the floating of point values, we can detect which node is attacked. In addition, the detection rate in our method is better than Error Back Propagation.