Matching Threshold Prediction Mechanism for Detector Maturation Based on GAs

摘要

Detector maturation plays an important role in network intrusion detection system. NSA (negative selection algorithm) which is designed for detector maturation has wide applications. And matching threshold r effects true positive and false positive of intrusion detection directly. In this paper a prediction calculation strategy of matching threshold r is designed, through which the optimum value of r is achieved. Because the static matching threshold r cannot meet the demand of dynamic network, self-adaptive matching threshold r based on GA is brought up to adjust r to changeable network intrusion. Simulation results prove that GA-based detector generation algorithm has better performances compared with NSA in aspect of true positive, false positive, and dynamic self-set condition.