AI-Driven Anomaly Detection in Cyber-Physical Systems: A Technical Approach to Real-Time Threat Mitigation

• Author(s): Apoorva Kasoju
• Paper ID: 1708037
• Page: 804-817
• Published Date: 31-10-2024
• Published In: Iconic Research And Engineering Journals
• Publisher: IRE Journals
• e-ISSN: 2456-8880
• Volume/Issue: Volume 8 Issue 4 October-2024

Abstract

As the core structure of contemporary infrastructure Cyber-Physical Systems (CPS) combine information technology through physical operation to unite calculations with networking functions and real systems. Safety-critical and real-time environments that use CPS reach a critical point where they need robust adaptive security mechanisms which can operate intelligently. Trusted security frameworks which utilize rules or signatures do not provide enough capabilities to identify new or sophisticated anomalies that occur within dynamic heterogeneous systems. The research presents an engineering-based approach for AI-based anomaly discovery in CPS that provides immediate threat countermeasures. This research conducts an assessment of deep autoencoders as well as convolutional neural networks (CNNs) and long short-term memory (LSTM) networks and reinforcement learning agents based on how each component can detect system deviations which signify potential cyber-attacks or faults within operational systems. The security framework uses a detection mechanism made from AI inference along with statistical analysis to achieve precision while cutting down on false positive errors. Our method receives validation by testing it with a simulated smart grid setup that runs different risk situations consisting of false data insertion and sensor manipulation and coordinated control deception. The framework determines performance measurements through detailed examination of detection accuracy together with latency amounts and scalability factors and resource consumption levels. The research confirms that this AI-driven strategy achieves superior results than conventional approaches alongside its automation for coping with emerging threats. This research provides evidence about integrating AI into CPS security loops and highlights important factors for real-time implementation in systems requiring low delay responses.

Keywords

AI-Driven Security, Anomaly Detection, Cyber-Physical Systems, Real-Time Threat Mitigation, Machine Learning, Smart Grid Security

Citations

IRE Journals:
Apoorva Kasoju "AI-Driven Anomaly Detection in Cyber-Physical Systems: A Technical Approach to Real-Time Threat Mitigation" Iconic Research And Engineering Journals Volume 8 Issue 4 2024 Page 804-817

IEEE:
Apoorva Kasoju "AI-Driven Anomaly Detection in Cyber-Physical Systems: A Technical Approach to Real-Time Threat Mitigation" Iconic Research And Engineering Journals, 8(4)