An efficient cyber-attack detection and classification in IoT networks with high-dimensional feature set using...

Area of Science:

• Cybersecurity
• Artificial Intelligence
• Network Security

Background:

• Internet of Things (IoT) networks face escalating cyber-attack challenges.
• Conventional security measures struggle with the speed and complexity of modern threats.
• There is a critical need for precise, efficient, and adaptive IoT security solutions.

Purpose of the Study:

• To propose a novel deep learning-based approach for enhanced cyber-attack detection in IoT networks.
• To evaluate the performance of the proposed model against traditional machine learning and deep learning techniques.
• To provide a scalable and robust solution for real-time identification of emerging cyber threats in IoT environments.

Main Methods:

• A deep learning approach utilizing feedforward neural networks.
• Optimization of the neural networks using the Levenberg-Marquardt algorithm.
• Comparative analysis against Support Vector Machines (SVM), Random Forest, and Artificial Neural Network (ANN) models.

Main Results:

• The proposed deep learning model achieved an accuracy rate of 99.7%.
• Exceptional performance metrics including precision, recall, and F1-score of 99.93%.
• Demonstrated minimal misclassifications and efficient processing of large data volumes for real-time detection.

Conclusions:

• The deep learning model significantly surpasses traditional methods in IoT cybersecurity.
• The system effectively reduces false positive rates and enhances attack classification accuracy.
• This research offers a scalable and robust solution for advancing cybersecurity in IoT environments.