Skip to Main Content
Artificial immune systems (AISs) are being increasingly utilized to develop network intrusion detection (NID) systems. The fundamental reason for their success in NID is their ability to learn normal behavior of a network system and then differentiate it from an anomalous behavior. As a result, they can detect a majority of innovative attacks. In comparison, classical signature based systems fail to detect innovative attacks. Light Weight Intrusion Detection System (LISYS) provides the basic framework for AIS based NID systems. This framework has been improved incrementally, including incorporation of thymus action, since it was first developed. In this paper, we have extended the basic thymus action model, which provides immature detectors with multiple chances to develop tolerization to normal. However, AIS is prone to successful attacks by malicious traffic which appears similar to the normal traffic. This results in high number of false positives. In this paper, we present a mathematical model of malicious traffic for TCP-SYN flood based distributed denial of services (DDoS) attacks. This model is used to generate different sets of malicious traffic. These sets are used for performance comparison of the proposed extended thymus action with the simple thymus action model. The results of our experiments demonstrate that the extended model has significantly reduced the number of false positives.