IDMM-IDS: An efficient and robust intrusion detection system for the IoT based on the inverted Dirichlet mixture...

Area of Science:

• Cybersecurity
• Network Security
• Internet of Things (IoT)

Background:

• The proliferation of Internet of Things (IoT) devices has created significant security vulnerabilities.
• Traditional intrusion detection systems (IDS) are often unsuitable for resource-constrained IoT environments due to high computational demands and poor adaptability.
• There is a critical need for efficient and robust IDS tailored for IoT security.

Purpose of the Study:

• To propose IDMM-IDS, an efficient and robust intrusion detection system specifically designed for Internet of Things (IoT) contexts.
• To address the challenges of computational overhead and adaptability in IoT intrusion detection.
• To enhance the detection of minority class threats in imbalanced datasets.

Main Methods:

• Utilized the inverted Dirichlet mixture model (IDMM) for modeling complex network traffic with minimal computational overhead.
• Employed extended stochastic variational inference (ESVI) for efficient model training and inference.
• Integrated a novel cluster-based oversampling technique to handle class imbalance issues.

Main Results:

• IDMM-IDS demonstrated superior detection performance compared to existing methods on the UNSW-NB15, WSN-DS, and WUSTL-IIOT-2021 datasets.
• The proposed system significantly reduced training and decision times, showcasing its efficiency.
• Effective detection of minority class threats was achieved without introducing noise into the dataset.

Conclusions:

• IDMM-IDS is a highly efficient and robust intrusion detection system suitable for resource-constrained IoT environments.
• The combination of IDMM and ESVI provides a powerful approach for analyzing IoT network traffic.
• The integrated oversampling technique effectively addresses class imbalance, improving threat detection capabilities.