OLM-CPS: Optimized Learning Models for the Security of Constrained Cyber Physical Systems

Machine learning is widely used for securing cyber physical systems, but their computational demands pose challenges in constrained environments. This paper reviews and analyzes optimization methods for learning models, including feature engineering, hyper-parameter optimization, and optimizer selection. It also proposes a methodology to utilize optimized learning models for cyber physical system security. By optimizing learning models, they become feasible for deployment in constrained systems. The paper explores various optimization techniques to enhance learning model efficiency and effectiveness. Feature engineering transforms raw data into meaningful representations, hyper-parameter optimization finds optimal configuration settings, optimizer selection chooses the most suitable algorithm, and federated learning reduces the node level computational requirements and also preserves privacy of individual nodes. The proposed methodology integrates optimized models into cyber physical systems to improve security, considering their resource limitations. This research contributes to the advancement of machine learning in securing cyber physical systems. The findings highlight the benefits of optimization methods and emphasize the importance of optimized learning models in constrained environments, enhancing security measures.