Design of an Intrusion Detection System for Unknown-attacks based on Bio-inspired Algorithms

Abstract

Signature-based Intrusion Detection System (IDS) can detect only known attacks that have signatures. As new unknown-attacks are appearing continuously, the detection of unknown-attacks has become the essential part of IDS. This paper presents a novel design of IDS by combining two existing bio-inspired machine learning algorithms; Artificial Immune System (AIS) and Ant Clustering Algorithm (ACA), and evaluates the pros and cons of the approach. In our approach, after ACA makes clusters by using unsupervised learning, AIS categorizes the network traffic to self and non-self as normal and abnormal profiles, respectively. Our design presents better performance than other existing similar design.