A Cooperative Intrusion Detection System for the Internet of Things Using Convolutional Neural Networks and Black...

Area of Science:

• Cybersecurity
• Network Security
• Artificial Intelligence

Background:

• Traditional intrusion detection systems (IDSs) are ineffective for Internet of Things (IoT) networks due to centralized designs.
• Increasingly complex infiltration techniques necessitate advanced security solutions for modern communication networks.

Purpose of the Study:

• To propose a novel, cooperative intrusion detection approach specifically designed for IoT environments.
• To enhance the accuracy and efficiency of threat detection in IoT communications.

Main Methods:

• Utilized Black Hole Optimization (BHO) for selecting critical communication attributes.
• Developed a matrix-based communication property description method.
• Implemented a Software-Defined Network (SDN) to partition networks into subnets, each monitored by a controller node.
• Employed a parallel combination of convolutional neural networks (PCNN) for threat identification within subnets.
• Integrated a majority voting mechanism for cooperative decision-making among controller nodes.

Main Results:

• The proposed cooperative strategy achieved high detection accuracy: 99.89% on the NSLKDD dataset and 97.72% on the NSW-NB15 dataset.
• Demonstrated a minimum improvement of 0.6% compared to previous intrusion detection approaches.
• Validated the effectiveness of BHO for feature selection and PCNN for threat analysis in IoT security.

Conclusions:

• The novel cooperative intrusion detection approach significantly improves security in IoT networks.
• The integration of BHO, SDN, PCNN, and majority voting offers a robust solution for detecting sophisticated network attacks.
• The findings highlight the potential of cooperative, AI-driven systems for future network security challenges.