An Intrusion Detection Model Based on Danger Theory for Wireless Sensor Networks

For the intrusion detection problem in Wireless Sensor Networks, an intrusion detection model based on the Danger Theory is proposed. The model has two layers including danger perception and control decision and detects intrusion by a multi-node cooperation mechanism. The model uses Danger Theory instead of SNS as the artificial immune theory basic, perceives dangers with Projection Pursuit Algorithm to handle the high dimension problem of network traffic information, classifies dangers with Extreme Learning Machine algorithm and uses Beta distribution trust evaluation to ensure the trust between nodes. By the simulations in the MATLAB with KDD CUP99 dataset, it is proved that the danger perception with Projection Pursuit Algorithm is effective, the classification speed of ELM algorithm is faster than SVM algorithm, the proposed model based on Danger Theory is better than the SNS model at the aspects of false negative rate, false positive rate and energy consumption.