End-to-End Network Traffic Examination for Network Intrusion Detection using Feature Embedding Learning
J. Arthy, Siddharth D Variar, N Tharakeshwar, Ritushree Narayan · 2024
The rapid evolution of information sharing and portable device technologies has led to a drastic increase in the number of smart devices, which helps in enhancing the level of comfort and intelligence in every household. It also drives the progress and innovation in different fields, including transportation and healthcare industries. Industrial Control Systems help in the development of society as a whole, by controlling and monitoring the flow of communication between the systems in a industrial network. ICSs have also advanced further due to the presence of smart devices. However, this has also led to a alarming increase in network security concerns. Data security is of the highest priority now. Attacks to the networks present in an ICS also lead to very dangerous outcomes, which may sometimes prove to be fatal. Anomaly detection techniques are mainly used in traditional intrusion detection systems. This is done by detecting any abnormal patterns in the network traffic. To develop an effective and flexible IDS, we are going to examine the Deep Neural Network in this paper. The continuous development of network behaviour and rapid increase in attacks call for development of IDS and evaluating many datasets formed over time using static and dynamic methods. This type of research leads to the identification of the most efficient algorithm for identifying any cyber-attacks in the future. The proposed model achieves higher accuracy than the previous machine learning models.