Enhancing Security in CPS Industry 5.0 using Lightweight MobileNetV3 with Adaptive Optimization Technique

Advanced Cyber-Physical Systems (CPS) that facilitate seamless communication between humans, machines, and objects are revolutionizing industrial automation as part of Industry 5.0, which is being driven by technologies such as IIoT, cloud computing, and artificial intelligence. In addition to providing flexible, individualized production processes, this growth brings with it fresh cybersecurity risks including Distributed Denial of Service (DDoS) attacks. This research suggests a deep learning-based approach designed to enhance security in CPS to address these issues. The system's primary goal is to identify and stop advanced cyberattacks. The strategy guarantees strong protection for industrial processes in a networked, intelligent environment. This study offers a sophisticated paradigm for improving Cyber-Physical Systems (CPS) security in Industry 5.0 by combining effective data preprocessing, thin edge computing, and strong encryption methods. The method starts with preprocessing the IoT23 dataset, which includes utilizing Gaussian filters to reduce noise, Mean Imputation to handle missing values, and Min-Max normalization to data scaling. The model uses flow-based, time-based, statistical, and deep feature extraction using ResNet-101 for feature extraction. Computational efficiency is maximized through the implementation of MobileNetV3, a thin convolutional neural network optimized for mobile and edge devices. The accuracy of the model is further improved by applying a Chaotic Tent-based Puma Optimization (CTPOA) technique. Finally, to ensure secure data transfer and protect private data in CPS settings, AES encryption is combined with discretionary access control. This comprehensive framework enables high performance, achieving 99.91% accuracy, and provides strong security for Industry 5.0 applications.