Anomaly Intrusion Detection Techniques: A Brief Review

In a broader sense detection of any unauthorized access of any information system is the basic aim of any intrusion detection system. However due to cost considerations it is practically impossible to provide total protection to an information system from intrusion for its entire useful life time. In this paper we provide a brief introduction to anomaly based intrusion detection systems that classify all reported techniques, including artificial immune systems (AIS), fuzzy logic (FL), swarm intelligence (SI), artificial neural networks (ANN), evolutionary computation (EC), and soft computing (SC). The various techniques of anomaly based intrusion detection system reported in the literature have been sorted out on the parameters like their strength and weakness. The important research contributions have been systematically compared and summarized to reflect the current status of research and challenges ahead. This will be helpful in knowing the new research directions. We also highlight the role of machine learning techniques for IDS. The contributions of research papers based on machine learning (ML) have also been considered. ML system have intrinsic properties like resilience to noisy data, adaptability, fault tolerance, robustness, low computational overhead etc, that provide a versatile tool in developing better intrusion detection techniques. We aim at providing a concise but comprehensive overview of research in progress and give direction to intrusion detection methods based on ensemble of ML techniques. This review work should be helpful and also provide critical insight into the current trend in IDS research especially in the application of ML approaches to IDS and related fields.