SCADA securing system using deep learning to prevent cyber infiltration

Area of Science:

• Computer Science
• Cybersecurity
• Industrial Control Systems

Background:

• Supervisory Control and Data Acquisition (SCADA) systems are critical for industrial operations but vulnerable to cyber-attacks due to constant internet connectivity and weak internal security.
• Traditional signature-based intrusion detection systems struggle to identify novel threats in complex SCADA environments.

Purpose of the Study:

• To propose and evaluate a novel intrusion detection algorithm, the Genetically Seeded Flora Transformer Neural Network (GSFTNN), to address SCADA system security vulnerabilities.
• To enhance the detection accuracy and efficiency of identifying cyber intrusions in industrial control systems.

Main Methods:

• The Genetically Seeded Flora (GSF) feature optimization algorithm is integrated with a Transformer Neural Network (TNN).
• The GSFTNN algorithm detects intrusions by identifying deviations in operational patterns indicative of unauthorized access.
• Performance evaluation was conducted using the WUSTL-IIOT-2018 ICS SCADA cybersecurity dataset.

Main Results:

• The proposed GSFTNN algorithm demonstrated superior performance compared to traditional methods like Residual Neural Networks (ResNet), Recurrent Neural Networks (RNN), and Long Short-Term Memory (LSTM).
• GSFTNN achieved higher accuracy and efficiency in detecting intrusions within the SCADA environment.
• The algorithm effectively identified subtle changes in operational patterns signaling potential cyber threats.

Conclusions:

• The GSFTNN algorithm offers a significant advancement in SCADA cybersecurity, providing a more robust and efficient solution for intrusion detection.
• This novel approach effectively mitigates security bottlenecks in industrial control systems.
• GSFTNN represents a promising alternative to conventional intrusion detection methods for protecting critical infrastructure.