The Industrial Revolution of technologies such as the Industrial Internet of Things (IIoT), cloud and Artificial Intelligence (AI) is breaking new frontiers in industrial process automation. Under the Industry 5.0 revolution, AI based manufacturing units are more sophisticated Cyber-Physical Systems (CPS) that allow the interaction of people, objects and machines at any given supply chain level. One of the key advantages of this transformation is that it enables implementing individual-focused and adaptable manufacturing systems. However, this interconnection poses various threats, especially the phenomenon referred to as attacks that are sophisticated in nature sometimes known as distributed denial of service (Ddos) attacks. In the quest to prevent cyber security challenges in the CPS Industry 5.0, this paper proposes a simple and effective System based on Deep Learning architecture. The very first stage concerns data acquisition which entails careful monitoring and collection of raw data from inbuilt sensors for real time performance. In this case, the data is also processed further in order to clean the data, handle missing values within the data set and fix any errors present in the data set, improving it. The next step involves Normalization and feature extraction which takes on the shape of reducing the data into shapes acceptable by the key features; flow-based, time-based and statistical features as well as deep features using ResNet-101. To process the models, MobileNetV3 which are light weight models of deep learning, are predicted to be utilized in the edge devices that are low in resources through quantization and pruning methods. This quantization and pruning is going to reduce the weight of the model or data. The efficient local search method CTPOA is applied for adjustment of parameters, data optimization and performance improvement of the model. Finally, the data in the CPS is safeguarded by the use of AES encryption and Discretionary Access Control policies.
