Detection of Node Replication Attacks in Wireless Sensor Networks Based on Combined Weighted k-Nearest Neighbor Classification
Wireless sensor network nodes have small size and strong concealment,so it is difficult to detect node replication attacks.The similarity between nodes is obtained according to the spatial position data of beacon nodes and the number of hops apart.The spatial co-ordinates of the horizontal axis and vertical axis of unknown nodes are solved by combining the Gaussian radial basis kernel function to determine the spatial position of each network node.The node replication attack model is established according to the attribute charac-teristics and voting mechanism of network nodes.The node types are divided by using the combined weighted k-nearest neighbor classifi-cation method,and the results are transmitted to the cluster head node,which makes the final arbitration to identify the node replication attack behavior.The simulation results show that the maximum detection rate of node replication attack is 99.5%and the minimum de-tection rate is 97.9%.The detection time of node replication attack is 5.41 s.The maximum number of communication overhead packets is 209 and the minimum number is 81.
国家科技期刊平台
NSTL
万方数据
