A Detector Generation Algorithm Based on Negative Selection

Detector generation is a crucial step of immune-based intrusion detection system. In order to improve the detecting efficiency of detectors and the dependence of matching threshold on an experienced value in the negative selection algorithm, an algorithm named VRGA for detector generation is proposed in this paper. Variable matching threshold r (r-variable) is introduced in VRGA to effectively increase the detector coverage. Clonal selection strategy considering detector similarity is adopted to increase the diversity of detectors. With above two improvements, VRGA brings the self-adaptability of matching threshold in negative selection and reduces the dependence of matching threshold on experience. In addition, VRGA increases the coverage and diversity of the detector set. The experimental results show that VGRA has better performance and its detecting efficiency is also improved correspondingly over the traditional negative selection algorithm.