Intrusion Detection System using Improved Wild Horse Optimizer-Based DenseNet for Cognitive Cyber-Physical System in Industry 4.0

A Cyber-Physical System (CPS) refers to a cyber and physical elements network which communicate with each other in a manner of feedback. A CPS is significant for daily activity and confirms crucial infrastructure as it generates the base for new smart devices. However, enhancing connectivity and CPS complexity establishes new security difficulties which makes Intrusion Detection Systems (IDS) a significant aspect of maintaining the reliability and integrity of these systems. Therefore, the Improved Wild Horse Optimizer-based DenseNet (IWHO-DenseNet) is proposed for IDS in CPS at Industry 4.0 using Deep Learning (DL). WHO is improved by using Random Running Strategy (RRS), Competition of Waterhole Mechanism (CWHM), and Dynamic Inertia Weight Strategy (DIWS) to enhance the WHO's performance and effectiveness. IWHO is established to select the most relevant features and DenseNet is employed to effectively detect and classify IDS in CPS. The existing techniques like Artificial Intelligence-Multimodal Fusion-based IDS (AIMMF-IDS), Quantum Dwarf Mongoose Optimization with Ensemble DL-based IDS (QDMO-EDLID), Stochastic Fractal Search Algorithm with DLIDS (SFSA-DLIDS), two-phase IDS, and Explainable AI-enabled IDS for Secure CPS (XAIID-SCPS) are employed to compare with IWHO-DenseNet. The IWHO-DenseNet achieves better accuracy of 99.81% compared to AIMMF-IDS, QDMO-EDLID, SFSA-DLIDS, two-phase IDS, and XAIID-SCPS respectively.