A Lightweight DNN-based IDS for Detecting IoT Cyberattacks in Edge Computing
Trong-Minh Hoang, Tuan-Anh Pham, Van-Viet Do, Van-Nhan Nguyen, Mạnh Hùng Nguyễn
Abstract
With the continuous growth of the Internet of Things applications, increasingly sophisticated and malicious network security attacks have been posing new security requirements. One of the first protection solutions to ensure security is to use an intrusion detection system (IDS) for detecting cyberattacks. Another hand, edge computing technology has been bringing many benefits to communication network infrastructure and IoT applications in terms of performance and privacy. However, the implementation of IDS systems on edge devices encounters many obstacles stemming from the resource constraints of edge devices. Hence, machine learning-based IDS systems have emerged to address such challenges. In this paper, we propose a lightweight deep neuron network-based IDS suitable for deployment at edge devices while still ensuring high attack detection accuracy. The evaluation results on the IoT23 dataset with various cases show that our proposed model has overcome previous proposals and reached an attack detection rate of 99%.