1. A Method of Choosing Important Features Used to Intrusion Detection System in Wireless Sensor Networks
- Author
-
Wang, Shu, Sha, Jia Xiang, Yu, Gui Xiang, Yu, Sheng Qin, and Yu, Sheng Chen
- Abstract
One of difficultest problem is to find these important features (namely, effective features) for intrusion detection system (IDS). In order to resolve the problem, a method is presented in the paper. The so-called important features are these features in which much information for IDS is provided. The information provided by a feature is measured by mean-squared error (namely, variance) of the feature. The correlativity between two features is measured by covariance. If the covariance is equal to zero, the two features are non-correlative. On condition that covariance is equal to zero, the new feature making the variance maximum is gained. The new feature is called first important features searched in the paper. In the same approach, the second, third … important features are gained. Tests show that the method developed in the article and IDS based on the important features sought in the method are useful and available.
- Published
- 2012
- Full Text
- View/download PDF