A Systematic Literature Review on Intrusion Detection Approaches

Resumen

Nowadays, intrusion detection systems play a major role in system security. Ideally, intrusion detection systems are capable of detecting intrusions in real time to prevent intruders from causing any harm to computer systems. Intrusion detection systems can be implemented using different intrusion detection approaches with its strengths and limitations. This paper provides an overview of the strengths and limitations of the different intrusion detection approaches, including Statistical-Based Anomaly, Pattern Matching, Data Mining and Machine Learning approach. The results show that Machine Learning is the most suitable approach for implementing intrusion detection system solutions, because of its ability to work as an automated process, which hardly needs human intervention. Using this partial conclusion, different machine learning techniques are studied and presented, also with their strengths and limitations. After the study, it can be concluded that the best technique to implement this kind of system is recurrent neural networks. An intrusion detection systems that hardly needs human intervention, can be developed and implemented, using this technique.