Detector Generation Algorithm Based on Online GA for Anomaly Detection

Abstract

T Detector plays an important role in intrusion detection system in artificial immune system, which makes detector generation algorithm especially significant. Traditional NSA cannot satisfy current network demands because the affinity limit r is difficult to fix in prior. A novel online GA-based algorithm is come up with self-adaptive mutation probability, in which affinity limit r is self-adaptive. Compared with GA-based detector maturation algorithm, detectors in online GA-based algorithm evolve online during the detection process which realizes self-organization and online learning to be adaptive to dynamic network. Finally simulation results testify that TP (true positive) value and FP (false positive) value of online GA-based algorithm is much better than NSA, GA-based and IGA-based algorithms without significant algorithm complexity increase.