Realguard: A Lightweight Network Intrusion Detection System for IoT Gateways.

Area of Science:

• Computer Science
• Cybersecurity
• Network Security

Background:

• The proliferation of Internet of Things (IoT) devices has led to significant cybersecurity challenges due to their inherent security flaws and limited defense mechanisms.
• Existing IoT gateways often lack robust security features, particularly deep learning-powered intrusion detection systems, due to high computational requirements.

Purpose of the Study:

• To introduce Realguard, a deep neural network (DNN)-based network intrusion detection system (NIDS) designed for direct operation on local IoT gateways.
• To provide real-time, accurate detection of multiple cyber attacks with a minimal computational footprint, addressing the limitations of current IoT security solutions.

Main Methods:

• Developed a lightweight feature extraction mechanism tailored for resource-constrained environments.
• Implemented an efficient deep neural network model for attack detection directly on IoT gateways.
• Evaluated Realguard's performance on practical datasets, including common IoT cyber threats.

Main Results:

• Realguard accurately detected ten types of cyber attacks, including port scans and Botnet, with an average accuracy of 99.57%.
• Achieved a superior detection rate compared to competitors, whose best performance was 98.85%.
• Demonstrated efficient operation on resource-constrained gateways like Raspberry Pi, processing approximately 10,600 packets per second.

Conclusions:

• Realguard offers a highly effective and efficient solution for real-time network intrusion detection in IoT environments.
• The system's lightweight design and high accuracy make it suitable for deployment on resource-constrained IoT gateways, significantly enhancing IoT security.
• Realguard successfully addresses the computational challenges of implementing deep learning-based security on edge devices.