A Comprehensive Review of Various Approaches to Intrusion Detection Systems

The extensive usage of the Internet to communicate and transfer data might succumb to various network related threats. Intrusions are one such threat, where the client/organization is at a risk of data theft. An intruder is someone who gains unauthorized access to our network or system. A network falling prey to an intrusion might result in loss of valuable data. A solution to intrusions is intrusion detection systems (IDS). This paper provides a comprehensive review of approaches to build IDS. The first section covers a review of the fundamentals of IDS, covering various intrusion types and IDSs, their strengths and their limitations. The next section discusses intrusions in wireless networks, followed by a review of a wireless approach to intrusion detection systems for IEEE 802.11 networks. The next section takes a look at various deep learning and machine learning approaches to intrusion detection systems that are currently being implemented. It summarizes the benchmark datasets that are currently being used to implement models for intrusion detection, followed by the results of a few machine learning models implemented on the NSL-KDD dataset.