Machine Learning-Driven Intrusion Detection for DDoS Attack Mitigation in Cyber-Physical Production Systems

The research aims to create an intelligent Intrusion Detection System (IDS) for Cyber-Physical Production Systems (CPPS) that uses machine learning approaches to identify Distributed Denial-of-Service (DDoS) attacks. The proposed approach trains and compares the performance of Random Forest (RF) and Deep Neural Networks (DNN). To train the various models, the dataset is first pre-processed by feature selection, normalisation, and splitting. Fast classification and interpretability are enabled by the RF model, while deep feature learning is used by the Deep Neural Networks model to identify intricate attack patterns. The Random Forest and Deep Neural Networks models achieved high accuracy scores of 98.2 and 99.3%, respectively, and low false positive rates, according to experimental assessments on benchmark datasets. These results show that the Deep Neural Networks based Intrusion Detection System is a good option for real-time industrial security applications as it effectively protects CPPS from changing cyberthreats.