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

Authors

  • Linlin Li College of software, University of Science and Technology Liaoning
  • Liangxu Sun
  • Gang Wang College of software, University of Science and Technology Liaoning

DOI:

https://doi.org/10.3991/ijoe.v14i09.8625

Keywords:

Intrusion Detection, Wireless Sensor Networks, Danger Theory, Extreme Learning Machine, Projection Pursuit, Beta Distribution

Abstract


This paper, due to the intrusion detection problem in Wireless Sensor Networks, proposes an intrusion detection model based on the Danger Theory instead of the traditional Self-NonSelf theory. The intrusion detection model has two layers structure including danger perception and control decision, and it uses a multi-node cooperation mechanism. The perception node can realize the danger perception with Projection Pursuit Algorithm, and the decision node can detect the intrusion in detail with Extreme Learning Machine Algorithm. The logic process between their layers is consistent with the Danger Theory. The proposed model can realize the data trust between nodes with the Beta distribution trust evaluation method. By the simulations in the MATLAB, the proposed intrusion detection model on the whole is better than the SNS model at the aspects including classification training, danger perception, false negative rate, false positive rate and energy consumption.

Downloads

Published

2018-09-30

How to Cite

Li, L., Sun, L., & Wang, G. (2018). An Intrusion Detection Model Based On Danger Theory for Wireless Sensor Networks. International Journal of Online and Biomedical Engineering (iJOE), 14(09), pp. 53–65. https://doi.org/10.3991/ijoe.v14i09.8625

Issue

Section

Papers