Intrusion Detection Model for Industrial Internet of Things Based on Improved Autoencoder

Area of Science:

• Computer Science
• Cybersecurity
• Network Engineering

Background:

• The increasing complexity of industrial Internet of Things (IoT) necessitates robust security measures.
• Ensuring the safety and controllability of industrial IoT systems is a growing concern.

Purpose of the Study:

• To propose an advanced detection method for enhancing the security of industrial IoT environments.
• To address the challenge of unbalanced network traffic data in security detection.

Main Methods:

• A detection method employing a stacked sparse autoencoder network model with simplified and sparse basic units.
• Utilizing a cascaded network structure to stack sparse autoencoder models for improved data handling.
• Incorporating a Softmax classifier for dynamic parameter adjustment and optimization of the detection model.

Main Results:

• The proposed method demonstrated excellent performance in identifying and detecting network attacks on the NSL-KDD dataset.
• Achieved a high accuracy index of approximately 95.42% for network attack detection.
• The detection time for the method was recorded at approximately 3.42 seconds.

Conclusions:

• The developed stacked sparse autoencoder network model effectively improves the security and detection capabilities for industrial IoT.
• The method offers a promising solution for real-time network attack identification in industrial settings.