A sustainable driven intrusion detection model for green CPS using ISP analysis and energy aware deep ensemble...

Area of Science:

• Cyber-Physical Systems (CPSs)
• Network Security
• Machine Learning

Background:

• Existing Intrusion Detection Systems (IDSs) in green CPSs lack the ability to detect Inconsistent Sequential Patterns (ISPs).
• This limitation leads to high misclassification rates and reduced system sustainability.
• There is a need for advanced IDS approaches that consider both security and energy efficiency in CPS environments.

Purpose of the Study:

• To propose a sustainability-driven framework for intrusion detection in green CPSs.
• To integrate Inconsistent Sequential Pattern (ISP) analysis with an energy-aware deep ensemble learning model.
• To enhance the reliability and sustainability of IDS in green CPSs.

Main Methods:

• Developed a multi-layered framework (perception, transport, network, control) for ISP-aware intrusion detection.
• Employed behavioral similarity modeling for fine-grained intrusion detection across CPS layers.
• Utilized an optimized deep ensemble learning model for efficient and generalized intrusion detection.

Main Results:

• Achieved high detection accuracies: 99.0654% for network data and 99.4523% for IoT data.
• Demonstrated significant enhancement in IDS reliability and resilience through ISP analysis.
• Validated the framework's effectiveness in a green CPS context.

Conclusions:

• The proposed ISP-aware intrusion detection framework significantly improves IDS performance in green CPSs.
• Integrating ISP analysis and energy-aware deep learning offers a sustainable solution for CPS security.
• The framework provides a robust and efficient method for detecting intrusions in complex CPS environments.