Detection of Network Intrusions using Hybrid Intelligent Systems

With the advancement in technology, our society has become so dependent on the internet and the number of internet users keeps rising each day. However, with the increased users comes the risk of hacking and malicious activities. One of the major concerns facing the technology sector today is the risk of intrusion. Thus, in the domain of security and computer networks, research in intrusion detection is essential. To combat the threats and malicious activities of the internet, the computer industry has gone a mile by creating new software and hardware products such as the Firewalls, Intrusion Prevention Systems and Detection Systems. Recently researchers created a Network-IDS to prevent such intrusions. However, these systems were prone to manipulations and had defects that were based on classification techniques. These systems failed to provide the necessary protections as they used a single classifier system or the individual classification technique. A single classifier classifies all of the data as normal or not, however due to the evolution of new attack patterns these systems failed to provide optimal attack detection rates with poor false alarm rates. The rise of different attack patterns meant that these systems cannot offer complete protection hence researchers came up with more sophisticated classification techniques that uses blends of several classification algorithms known as a hybrid intelligent system, leading to more detection accuracy. The aim of this study is to contrast various classifiers for network intrusions while combining these classifiers to direct the study towards hybrid intelligent systems. The study is carried out by performing an empirical and literature review while simultaneously providing a base for future studies.