Designing Abstract Immune Mobile Agents for Distributed Intrusion Detection

Inspired by the self-protection mechanism of biological immune system which can be regarded as a typical multi-agent system, a novel distributed intrusion detection model based on immune mobile agents is proposed in this paper. In this model, every immune agent owns a small detector set which represents its detection function. After biological immune principles are introduced, the corresponding distributed intrusion detection model is described in detail. With the detector set generated by evolutionary negative selection algorithm, the immune agents can evolve to adapt to the change of intrusion patterns according to the real circumstance. Therefore the anomaly changes in large non-self space can be detected by the immune agents with relatively small detector sets. This paper focuses on the immune agents´ initialization, detection, evolution and update strategies. The immune agents are abstract and their detailed implementations are not taken into account in this paper. Finally the characteristics of this distributed model are also discussed in detail