Research Of Detector Generation Algorithm Based On Negative Selection

With the rapid development of the Internet, network security has become the focus which is paid more attention to. The traditional network security techniques can hardly deal with kinds of network attack, as an initiative technique of information security, the intrusion detection system remedies effectively the disadvantage of traditional safety technique. Because intrusion detection system draw lesson from distribution, diversity, adaptability of the biology immune system, the research on intrusion detection system based on biology immune principle becomes the hotspot of network security in recent years.Although the research of intrusion detection system based on immune principle has been developed greatly, there are still some problems on it. The low detecting rate to the mutation of known and unknown intrusion is the main problem of intrusion detection system. The detecting rate of intrusion system is mainly decided by the effectivity of detectors, the negative selection algorithm is mainly adopted by generation of system detector. But the detector generated by the algorithm has a lot of superposition, which influences seriously the overlay space of the detector and reduces the detection rate and the detectors with much redundancy also have bad effect on the diversity of intrusion detection system, and reduce the detection rate for unknown intrusion. In this dissertation, a new intrusion detection model is presented for the above questions. The model of detectors generation is analyzed and the former negative selection algorithm is improved. The overlap detectors which are selected by biology niche strategy are processed with mutation with the improved algorithm. Detectors with high match value are set to low frequency mutation, while the ones with low match value are set to high frequency mutation. The low frequency mutation and high frequency mutation can reduce the superposition of detectors and enhance their diversity. Finally, the experiment proves that improved negative selection algorithm can increase the detecting rate of intrusion detection system.