Hybrid Genetic Fuzzy Inference Engine to Detect Intrusion in Networks

Abstract

With the drastic increase in internet usage, various categories of attacks have also evolved. These attacks exploit system vulnerabilities thus posing danger to the sensitive or private information stored in the system thus violating confidentiality, integrity and availability of the resources. Conventional intrusion detection techniques like firewall have been implemented to counter these attacks but they too have failed due to the increased potential of the attackers or hackers as they use innovative ideas to attack the system. Thus substantial systems are needed to eliminate these attacks before they inflict huge damage to the organization by retrieving all the sensitive information. Computational techniques are currently being considered as a novel field to detect intrusions due to their characteristic properties such as adaptability, fault tolerance and higher accuracy. Genetic fuzzy system is considered to be the most suitable approach for constructing the intrusion detection system. The proposed approach is the hybrid of fuzzy logic with genetic algorithm and then implying mathematical model to cover the whole dataset, thus making the approach suitable for high dimensional problems. The fuzzy logic constructs precise and flexible patterns while the genetic algorithm based on evolutionary computation helps in attaining an optimal solution due to its learning capability. The proposed approach has been applied on KDD-99 dataset, which is the most widely used network dataset and is able to classify normal connections as well as anomalies with greater precision. The intended approach has been compared with the existing genetic fuzzy systems used in intrusion detection and other approaches on basis of various metrics to check its performance and conciseness in classification.