Anomaly-Based Network Intrusion Detection Model Using Deep Learning in Airports

Abstract

The number of cyber-attacks are growing quickly and we are encountering modern and complex network intrusion attacks everyday even in secure computer networks. Last year, many airports in different countries were under attack of multiple network intrusions in various cyber-segments especially Information and Communication Technology (ICT) system (e.g. Ransomware attacks). Such cyber-attacks could happen again in much more destructive ways which can cause irreparable losses, and endanger human life by disruption and corruption of the airport ICT system. We are approaching an anomaly-based Network Intrusion Detection System (IDS) using deep learning which provides a normal system behavior model and detects an abnormal behavior. In other words, this model is designed to detect not only known network intrusion attacks, but also unknown and modern attacks. We have trained and tested our model with DARPA dataset used in KDD 1999 Cup. Our model achieved an outstanding result with highly accurate detection rate, also low false alarm rate, which is superior to the previous researches conducted on this dataset.