Deep Learning-Enabled Threat Intelligence Scheme in the Internet of Things Networks

With the prevalence of Internet of Things (IoT) systems, there should be a resilient connection between Space, Air, Ground, and Sea (SAGS) networks to offer automated services to end-users and organizations. However, such networks suffer from serious security and safety issues if IoT systems are not protected efficiently. Threat Intelligence (TI) has become a powerful security technique to understand cyber-attacks using artificial intelligence models that can automatically safeguard SAGS networks. In this paper, we propose a new TI scheme based on deep learning techniques that can discover cyber threats from SAGS networks. The proposed scheme contains three modules: a deep pattern extractor, TI-driven detection and TI-attack type identification technique. The deep pattern extractor module is designed to elicit hidden patterns of IoT networks, and its output used as input to the TI-driven detection. TI-attack type identification is used to identify the attack types of malicious patterns to assist in responding to security incidents. The proposed scheme is evaluated on the two datasets of TON-IoT and N-BAIOT. The experimental results prove that the scheme achieves high performances in terms of the detection and false alarm rates compared with other similar techniques.

Automated summary

This paper introduces a new threat intelligence scheme based on deep learning techniques, which can identify cyber threats from SAGS networks and has shown high performance in terms of detection and false alarm rates in experiments.