A lightweight framework to secure IoT devices with limited resources in cloud environments.

Area of Science:

• Cybersecurity
• Machine Learning
• Internet of Things (IoT)

Background:

• Billions of Internet of Things (IoT) devices are vulnerable to cyber threats due to limited resources.
• IoT devices often act as gateways to cloud infrastructures, increasing their attack surface.
• Resource constraints in IoT hinder the implementation of robust security measures.

Purpose of the Study:

• To propose a lightweight intrusion detection framework for resource-constrained IoT environments.
• To enable real-time anomaly detection in IoT networks.
• To enhance IoT security without compromising device performance.

Main Methods:

• Developed a lightweight decision tree-based intrusion detection framework.
• Implemented a novel leaf-cut feature optimization strategy.
• Utilized adaptive cloud-edge intelligence for efficient processing.

Main Results:

• Achieved high accuracy (98.2% on NSL-KDD, 97.9% on Bot-IoT) with less than 1% false positives.
• Demonstrated significant improvements over traditional models (e.g., SVM, Neural Networks) in accuracy and energy efficiency (up to 78% less energy).
• Real-time inference achieved in under 1 ms on Raspberry Pi nodes at 1,250 samples/sec with only 12.5 MB memory usage.

Conclusions:

• The proposed framework offers an energy-efficient, scalable, and interpretable solution for IoT security.
• Suitable for real-time anomaly detection in diverse IoT applications like smart cities and healthcare.
• Addresses the critical need for effective security in resource-limited IoT ecosystems.