ERID: A Deep Learning-based Approach Towards Efficient Real-Time Intrusion Detection for IoT
Murao Lin, Baokang Zhao, Qin Xin · 2020
In the 5G and Internet of Things (IoT) era, the threat of network intrusions has greatly affected people's work and life. The increasing complexity of intelligent devices in IoT brings huge challenges to the network intrusion detection. We address these issues and propose a novel intrusion detection system (IDS) called ERID, which is based on a real-time anomaly detection approach. A new type of unsupervised stacked auto-encoders (SAE), is trained by using normal network traffics, then is assessed on its ability to detect four types of attack in IoT. In this work, we mainly focus on the classification of normal and threat patterns. The extensive experimental results based on a famous real-world dataset have been conducted, which also demonstrated the effectiveness and superiority of our novel ERID scheme compared with the existing works in the literature. We hope our work can be used to further stimulate real-time intrusion detection approaches for IoT.