A New Intrusion Detection System for the Internet of Things via Deep Convolutional Neural Network and Feature...

Area of Science:

• Computer Science
• Cybersecurity
• Network Engineering

Background:

• The Internet of Things (IoT) is prevalent in automated systems, collecting sensitive data that attackers target.
• Existing intrusion detection systems (IDS) struggle with performance and classifying diverse cyberattacks.
• There is a need for advanced IDS to secure IoT environments.

Purpose of the Study:

• To propose a Deep Convolutional Neural Network (DCNN)-based IDS for enhanced IoT security.
• To improve IDS performance and reduce computational power requirements.
• To effectively identify subcategories of cyberattacks in IoT networks.

Main Methods:

• A DCNN model comprising two convolutional and three fully connected dense layers was developed.
• The model was trained and evaluated using the IoTID20 dataset.
• Optimization techniques including Adam, AdaMax, and Nadam were applied.

Main Results:

• The proposed DCNN-based IDS demonstrated high accuracy, precision, recall, and F1-score.
• Optimization techniques significantly enhanced model performance.
• The DCNN model outperformed existing deep learning and traditional machine learning techniques.

Conclusions:

• The developed DCNN-based IDS offers a robust and accurate solution for IoT cybersecurity.
• The model effectively addresses the limitations of current IDS in performance and attack classification.
• This approach represents a significant advancement in securing the Internet of Things.