Point-Wise Activations and Steerable Convolutional Networks for DDoS-Attack Detection in Cyber-Physical Systems Over 5G Networks

Abstract

The growth in DDoS attacks in CPS over 5G networks has emerged as the major risks affecting the reliability and continuity of car supply chain systems. Old school approaches to detection fail to work properly within 5G environments because of large and constantly changing volumes of traffic data that cannot be easily filtered for malicious patterns. In order to overcome these problems, this research work suggests a new framework that combines Point-Wise Activations with Steerable Convolutional Networks (PSCNs) with Circulatory System-Based Optimization (CSBO) for DDoS attack detection. The PSCNs excel in extracting both global and local information from network traffic, while the CSBO is tasked with optimizing the hyperparameters and weights of the network, thereby enhancing its performance. The current method proficiently addresses the issue and achieves an accuracy of 99.9% in comparison to other heuristics. Consequently, the CSBO, which employs adaptive and efficient optimization, ensures that the proposed framework delivers highly accurate real-time DDoS detection methods and is dependable for enhancing security in both current and future 5G-enabled Cyber-Physical Systems (CPS).