An intrusion detection system based on data mining and immune principles

In this paper, a framework of an immune-based intrusion detection system (IDS) is presented. Here data mining techniques are used to discover frequently occurring patterns, which are equivalent to self proteins in the immune system. During the tolerance process known as negative selection, a set of valid detectors that does not match any self protein mined previously is generated in the space of nonself based on a distance metric. These negative detectors are distributed into the network system to perform anomaly detection independently and concurrently. Our experiment confirms a low false positive rate and a high detection rate.